WO2023218957A1 - Information processing device, information processing method, and computer program - Google Patents

Information processing device, information processing method, and computer program Download PDF

Info

Publication number
WO2023218957A1
WO2023218957A1 PCT/JP2023/016397 JP2023016397W WO2023218957A1 WO 2023218957 A1 WO2023218957 A1 WO 2023218957A1 JP 2023016397 W JP2023016397 W JP 2023016397W WO 2023218957 A1 WO2023218957 A1 WO 2023218957A1
Authority
WO
WIPO (PCT)
Prior art keywords
sodium concentration
serum sodium
information processing
subject
future
Prior art date
Application number
PCT/JP2023/016397
Other languages
French (fr)
Japanese (ja)
Inventor
慎太郎 大山
珠希 木下
寛 有馬
大輔 萩原
Original Assignee
国立大学法人東海国立大学機構
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 国立大学法人東海国立大学機構 filed Critical 国立大学法人東海国立大学機構
Publication of WO2023218957A1 publication Critical patent/WO2023218957A1/en

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B10/00—Other methods or instruments for diagnosis, e.g. instruments for taking a cell sample, for biopsy, for vaccination diagnosis; Sex determination; Ovulation-period determination; Throat striking implements
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H20/00—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/50—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for simulation or modelling of medical disorders
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16Y—INFORMATION AND COMMUNICATION TECHNOLOGY SPECIALLY ADAPTED FOR THE INTERNET OF THINGS [IoT]
    • G16Y10/00—Economic sectors
    • G16Y10/60—Healthcare; Welfare
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16Y—INFORMATION AND COMMUNICATION TECHNOLOGY SPECIALLY ADAPTED FOR THE INTERNET OF THINGS [IoT]
    • G16Y20/00—Information sensed or collected by the things
    • G16Y20/40—Information sensed or collected by the things relating to personal data, e.g. biometric data, records or preferences

Definitions

  • the technology disclosed herein relates to an information processing device and the like for predicting serum sodium concentration at a future time.
  • Hyponatremia is an electrolyte abnormality in the body in which the serum sodium concentration is lower than a predetermined value (eg, 135 mEq/L), and is accompanied by symptoms such as nausea, headache, and light-headedness.
  • Severe hyponatremia eg, serum sodium concentration below 125 mEq/L
  • can cause severe central nervous system symptoms such as impaired consciousness, convulsions, and coma.
  • hyponatremia can be categorized into three types: (1) a decrease in extracellular fluid volume, (2) an almost normal extracellular fluid volume, and (3) an increase in extracellular fluid volume. It is broadly divided into Hyponatremia associated with decreased extracellular fluid loss can be treated by replenishing extracellular fluid by administering intravenous fluids (e.g., physiological saline, hypertonic (3%) saline, No. 3 solution, glucose solution, etc.). It will be done.
  • intravenous fluids e.g., physiological saline, hypertonic (3%) saline, No. 3 solution, glucose solution, etc.
  • treatments for hyponatremia with a nearly normal or increased extracellular fluid volume include water restriction and promotion of excretion of free water stored in the body using vasopressin V2 receptor antagonists. . Thus, the treatments for both are diametrically opposed.
  • ODS Osmotic Demyelination Syndrome
  • Patent Document 1 Conventionally, systems for predicting various medical events from information in electronic medical records have been proposed (see, for example, Patent Document 1).
  • the conventional system described above does not predict serum sodium concentration at a future time.
  • a technology that accurately predicts serum sodium concentration at a future time is desired.
  • This specification discloses a technique that can solve the above-mentioned problems.
  • the information processing device disclosed in this specification is a device for predicting serum sodium concentration at a future time, and includes a prediction factor acquisition unit and a prediction execution unit.
  • the predictive factor acquisition unit includes, for the subject of the prediction, a measured value of serum sodium concentration at a reference time, and an index value representing sodium intake in a future period that is a period from the reference time to the future time. Get predictors.
  • the prediction execution unit applies the serum sodium concentration prediction model generated by machine learning using training data in which the predictive factors and the measured value of serum sodium concentration at the future time are associated with the serum sodium concentration obtained for the subject. By inputting the predictive factor, the subject's serum sodium concentration at the future time is predicted.
  • this information processing device uses predictive factors that include the measured value of serum sodium concentration at the reference time and the index value representing sodium intake in the future period for the subject of prediction to predict serum sodium concentration. By inputting it into the model, it is possible to accurately predict the subject's serum sodium concentration at a future time.
  • the information processing apparatus may further include a prediction result output unit that outputs a prediction result of the subject's serum sodium concentration.
  • the information processing device further includes a training data acquisition unit that acquires the training data, and a model acquisition unit that creates the serum sodium concentration prediction model by the machine learning using the training data. You can also use it as If this configuration is adopted, a serum sodium concentration prediction model can be obtained without using any other device, and the serum sodium concentration of the subject at a future time can be predicted using this model.
  • the machine further uses the updated training data including data in which the predictive factor for the subject is associated with a measured value of serum sodium concentration at the future time.
  • the configuration may include a model updating section that updates the serum sodium concentration prediction model through learning. If this configuration is adopted, the serum sodium concentration prediction model can be made into a model that is more suitable for the characteristics of the subject, and the prediction accuracy of the serum sodium concentration can be improved.
  • the predictive factor acquisition unit acquires information specifying the type of ingestion that the subject ingests, and also refers to information indicating the amount of sodium contained in each of the ingestions. Then, an index value representing the amount of sodium intake in the future period may be acquired.
  • the user can obtain an index value representing sodium intake in the future period simply by specifying the type of food consumed, and more efficiently predict serum sodium concentration. can do.
  • the predictor further includes an index value representing the amount of water intake in the future period, and an index value representing the amount of water discharged in the past period, which is a period from the past time to the reference time. It is also possible to have a configuration including the following. If this configuration is adopted, the accuracy of predicting serum sodium concentration can be improved.
  • the subject may be a hyponatremic patient.
  • a rate of increase in concentration from a measured value of the serum sodium concentration of the subject at the reference time to a predicted value of the serum sodium concentration of the subject at the future time is further set in advance.
  • the configuration may include a notification section that provides notification using a predetermined method when the speed is faster than a threshold value.
  • the technology disclosed in this specification can be realized in various forms, such as an information processing device, an information processing method, a computer program that implements these methods, and a temporary computer program that records the computer program. It can be realized in the form of a non-standard recording medium or the like.
  • Explanatory diagram conceptually showing the serum sodium concentration prediction model MO
  • Explanatory diagram showing the configuration of the hospital system 10 A block diagram schematically showing the configuration of a terminal device 100
  • Explanatory diagram showing an example of infusion information ID Flowchart showing the serum sodium concentration prediction model acquisition process
  • Flowchart showing serum sodium concentration prediction process Explanatory diagram showing the prediction accuracy of the serum sodium concentration prediction model MO of this example
  • Explanatory diagram showing the prediction accuracy of the serum sodium concentration prediction model MO when the learning model and predictive factors are variously changed
  • FIG. 1 is an explanatory diagram conceptually showing a serum sodium concentration prediction model MO.
  • the serum sodium concentration prediction model MO is a trained model used to predict the serum sodium concentration at a future time.
  • the serum sodium concentration prediction model MO is a machine learning method that uses seven predictors as input and outputs (response) the serum sodium concentration SSC (t+ ⁇ t) (mEq/L) at future time (t+ ⁇ t). It's a model.
  • the predictive factors as input include the following seven items.
  • the reference time t is a time that serves as a reference for predicting the serum sodium concentration SSC(t+ ⁇ t) at a future time (t+ ⁇ t), and is a time at which the measured value of the serum sodium concentration SSC(t) is known.
  • the reference time t can be any arbitrary time, but is, for example, the time at which the serum sodium concentration is predicted using the serum sodium concentration prediction model MO.
  • the time interval between the reference time t and the future time (t+ ⁇ t), and the time interval between the past time (t ⁇ t) and the reference time t, ⁇ t can take any value, but in this embodiment, It is set to a fixed value of 6 hours.
  • ⁇ t may be a variable value.
  • the future time (t+ ⁇ t) may be any future time seen from the reference time t, and is not necessarily limited to an actual "future" time.
  • the reference time t is a time that is ( ⁇ t ⁇ 2) from the present
  • a time that is ⁇ t back from the present corresponds to “future time (t+ ⁇ t)” as viewed from the reference time t.
  • infusion dose IV in future period T (0) from reference time t to future time (t + ⁇ t)" and/or “sodium amount ISC contained in infusion” is an index value representing the sodium intake at T(0).
  • the “infusion dose IV in the future period T(0)” is an index value representing the water intake amount in the future period T(0).
  • the urine volume UV in the past period T(-1) from the past time (t- ⁇ t) to the reference time t is an index value representing the amount of water excreted in the past period T(-1).
  • the predictive factors as inputs of the serum sodium concentration prediction model MO are the measured value of the serum sodium concentration at the reference time t and the future period T(0) from the reference time t to the future time (t+ ⁇ t).
  • FIG. 2 is an explanatory diagram showing the configuration of the in-hospital system 10.
  • the in-hospital system 10 is an information system introduced within a hospital.
  • the in-hospital system 10 of this embodiment uses the above-described serum sodium concentration prediction model MO to predict the serum sodium concentration at a future time when a predetermined treatment is given to the patient. executed.
  • the in-hospital system 10 includes a terminal device 100 used by a medical worker P1 such as a doctor or nurse, and a server device 200 installed within the hospital.
  • the devices constituting the in-hospital system 10 are communicably connected to each other via the communication network NET.
  • the terminal device 100 is, for example, a PC, a tablet terminal, a smartphone, or the like.
  • FIG. 3 is a block diagram schematically showing the configuration of the terminal device 100.
  • the terminal device 100 includes a control section 110, a storage section 120, a display section 130, an operation input section 140, and an interface section 150. These units are communicably connected to each other via a bus 190.
  • the terminal device 100 is an example of an information processing device within the scope of the claims.
  • the display unit 130 of the terminal device 100 is composed of, for example, a liquid crystal display, and displays various images and information.
  • the operation input unit 140 includes, for example, a keyboard, a mouse, buttons, a microphone, etc., and receives operations and instructions from the medical worker P1. Note that the display unit 130 may function as the operation input unit 140 by including a touch panel.
  • the interface section 150 is configured with, for example, a LAN interface, a USB interface, etc., and communicates with other devices by wire or wirelessly.
  • the storage unit 120 of the terminal device 100 is composed of, for example, ROM, RAM, hard disk drive (HDD), solid state drive (SSD), etc., and stores various programs and data, and performs work when executing various programs. It is used as a temporary storage area for space and data.
  • the storage unit 120 stores a serum sodium concentration prediction program CP, which is a computer program for executing various processes described below.
  • the serum sodium concentration prediction program CP is provided, for example, in a state stored in a computer-readable recording medium (not shown) such as a CD-ROM, DVD-ROM, or USB memory, or is provided externally via the interface section 150. It is provided in a state that can be obtained from a device (for example, a server on a cloud or another terminal device), and is stored in the storage unit 120 in a state that can be operated on the terminal device 100.
  • the storage unit 120 of the terminal device 100 stores training data TD, serum sodium concentration prediction model MO, predictive factor data PD, and prediction result data RD in various processes described below. These data and models will be explained in conjunction with explanations of various processes later.
  • the control unit 110 of the terminal device 100 is configured by, for example, a CPU, and controls the operation of the terminal device 100 by executing a computer program read from the storage unit 120.
  • the control unit 110 reads the serum sodium concentration prediction program CP from the storage unit 120 and executes it, thereby functioning as a serum sodium concentration prediction processing unit 111 that executes various processes described below.
  • the serum sodium concentration prediction processing section 111 includes a training data acquisition section 112, a model acquisition section 113, a predictive factor acquisition section 114, a prediction execution section 115, a prediction result output section 116, a model update section 117, and a notification section. 118. The functions of these units will be explained in conjunction with explanations of various processes later.
  • the server device 200 (FIG. 2) is a device that provides an information presentation function and an information processing function in the in-hospital system 10.
  • the server device 200 stores infusion information ID.
  • FIG. 4 is an explanatory diagram showing an example of infusion information ID.
  • the infusion information ID is information that associates the type (preparation) of an infusion to be administered to a hyponatremic patient with the amount of sodium and potassium (mEq/L) contained in each type of infusion. By referring to the infusion information ID, the amounts of sodium and potassium contained in each type of infusion can be specified.
  • FIG. 5 is a flowchart showing the serum sodium concentration prediction model acquisition process.
  • the serum sodium concentration prediction model acquisition process is a process for acquiring the serum sodium concentration prediction model MO mentioned above.
  • the terminal device 100 acquires the serum sodium concentration prediction model MO by creating the serum sodium concentration prediction model MO by itself using predetermined machine learning.
  • the serum sodium concentration prediction model acquisition process is started in response to the medical worker P1 operating the operation input unit 140 of the terminal device 100 to input a start instruction.
  • the training data acquisition unit 112 (FIG. 3) of the terminal device 100 acquires training data TD (S110).
  • the training data TD is a set of multiple data in which the seven predictive factors shown in FIG. 1 are associated with the measured value (actual value) of the serum sodium concentration SSC (t+ ⁇ t) at a future time (t+ ⁇ t).
  • Training data TD is acquired via the interface section 150 and stored in the storage section 120.
  • the model acquisition unit 113 (FIG. 3) of the terminal device 100 creates a serum sodium concentration prediction model MO by predetermined machine learning (including deep learning) using the training data TD (S120).
  • the model acquisition unit 113 uses the seven predictive factors included in the training data TD as explanatory variables, uses the measured value of serum sodium concentration SSC (t+ ⁇ t) at a future time (t+ ⁇ t) included in the training data TD as an objective variable, and uses a predetermined
  • a serum sodium concentration prediction model is created by performing machine learning based on a predetermined learning algorithm while referring to evaluation indicators (e.g. root mean square error (RMSE), mean absolute error (MAE), coefficient of determination (R 2 )). Create MO.
  • evaluation indicators e.g. root mean square error (RMSE), mean absolute error (MAE), coefficient of determination (R 2 )
  • the created serum sodium concentration prediction model MO is stored in the storage unit 120 of the terminal device 100. Through the above steps, the serum sodium concentration prediction model acquisition process is completed.
  • FIG. 6 is a flowchart showing the serum sodium concentration prediction process.
  • the serum sodium concentration prediction process is a process for predicting the serum sodium concentration at a future time. More specifically, the serum sodium concentration prediction process is based on the serum sodium concentration SSC (t+ ⁇ t) at a future time (t+ ⁇ t) when a predetermined treatment is given to the patient with hyponatremia. This is a process of predicting using the concentration prediction model MO.
  • the serum sodium concentration prediction process is started in response to the medical worker P1 operating the operation input unit 140 of the terminal device 100 and inputting a start instruction.
  • the predictive factor acquisition unit 114 (FIG. 3) of the terminal device 100 acquires the seven predictive factors (FIG. 1) described above for a hyponatremic patient who is a prediction target (S310).
  • the acquired predictive factors are stored in the storage unit 120 as predictive factor data PD.
  • the medical worker P1 conducts a blood test on the patient and inputs the test results (each measured value) via the operation input unit 140. input.
  • the predictive factor acquisition unit 114 acquires each input measurement value.
  • these predictors may be obtained by other means (eg, using PoCT or an implantable device).
  • values inferred from analysis results of body fluids for example, tears and sweat may be acquired.
  • the medical worker P1 determines the type and dose of the infusion to be administered to the patient, and inputs the determination results (type and dose of the infusion) via the operation input unit 140. do.
  • the predictive factor acquisition unit 114 acquires information specifying the type and dosage of the input infusion, and also refers to the infusion information ID (FIG. 4) stored in the server device 200 to determine the amount of sodium contained in the infusion. Obtain ISC (mEq/L) and potassium amount IPC (mEq/L).
  • the medical worker P1 measures the patient's urine volume UV in the past period T(-1) and inputs the measurement result (urine volume UV) via the operation input unit 140. do.
  • the predictive factor acquisition unit 114 acquires the input urine volume UV. ⁇ Urine volume UV (ml) in the past period T (-1) from past time (t- ⁇ t) to reference time t
  • the predictive factor acquisition unit 114 may acquire predictive factors from an electronic medical record provided in the in-hospital system 10 or an external network.
  • the serum sodium concentration prediction processing unit 111 (FIG. 3) of the terminal device 100 determines whether or not it is the first loop for the patient to be predicted (S320), and if it is the first loop (S320: YES), the process skips the process of S330, which will be described later, and proceeds to the process of S340.
  • the prediction execution unit 115 (FIG. 3) of the terminal device 100 inputs into the serum sodium concentration prediction model MO the predictive factors acquired for the patient to be predicted, thereby predicting the target patient at future time (t+ ⁇ t).
  • the serum sodium concentration SSC (t+ ⁇ t) is predicted (S340).
  • the prediction execution unit 115 generates prediction result data RD, which is information indicating the prediction result of the subject's serum sodium concentration SSC (t+ ⁇ t) at a future time (t+ ⁇ t), and stores it in the storage unit 120 of the terminal device 100.
  • the prediction result output unit 116 of the terminal device 100 outputs the prediction result of the subject's serum sodium concentration SSC (t+ ⁇ t) at future time (t+ ⁇ t) based on the prediction result data RD (S350). For example, the prediction result output unit 116 causes the display unit 130 to display the prediction result. As a result, the medical worker P1 can grasp the predicted value of the serum sodium concentration SSC (t+ ⁇ t) at a future time (t+ ⁇ t) when a specific type of infusion is administered at a specific dose to the target patient. can do.
  • the notification unit 118 (FIG. 3) of the terminal device 100 determines the predicted rate of increase in the serum sodium concentration of the target patient (i.e., from the measured value of the serum sodium concentration SSC(t) at the reference time t). If the rate of increase in serum sodium concentration SSC (t+ ⁇ t) to the predicted value at future time (t+ ⁇ t) is faster than a preset threshold (upper limit), a notification process is performed using a predetermined method (S350).
  • the threshold value for the rate of increase in serum sodium concentration at this time is set, for example, to 8 to 10 mEq/L/day in order to avoid ODS (osmotic demyelination syndrome).
  • examples of methods of notification processing include displaying an alarm image on the display unit 130, outputting an alarm sound using an audio output means (not shown), and the like.
  • the medical worker P1 can recognize that the predicted increase rate of the serum sodium concentration is too fast and may cause ODS.
  • the medical worker P1 changes the type and/or dose of the infusion to be administered to the target patient so that the predicted rate of increase in serum sodium concentration will be slower, and The device 100 is caused to execute the processes of S310 to S350 again.
  • the healthcare worker P1 determines the type and/or dosage of the infusion that allows hyponatremia to be treated as quickly as possible while avoiding triggering ODS. be able to.
  • the serum sodium concentration prediction processing unit 111 of the terminal device 100 performs the process after S310 described above. Execute the same process. In the second and subsequent loops, the reference time t is updated to the future time (t+ ⁇ t) in the previous loop. That is, for example, in S310, the predictive factor acquisition unit 114 (FIG. 3) acquires the following seven predictive factors.
  • the model update unit 117 (FIG. 3) of the terminal device 100 uses the predictive factors in the previous loop and the reference time t in the current loop (in the future in the previous loop).
  • the serum sodium concentration prediction model MO is updated by machine learning using training data TD including data associated with the measured value of serum sodium concentration SSC(t) at time (t+ ⁇ t) (S330).
  • the serum sodium concentration prediction model MO becomes a model that is more suitable for the characteristics (pathological condition, constitution, etc.) of the target patient, and the prediction accuracy of the serum sodium concentration is improved.
  • serum sodium concentration prediction model MO An example of the above-mentioned serum sodium concentration prediction model MO will be described below.
  • serum sodium concentration was predicted by machine learning using data obtained during the treatment process of hyponatremic patients (16 cases) admitted to the Department of Diabetes and Endocrinology, Nagoya University Hospital.
  • a model MO was created.
  • FIG. 7 is an explanatory diagram showing the prediction accuracy of the serum sodium concentration prediction model MO of this example.
  • FIG. 7 shows the relationship between the measured values of serum sodium concentration and the predicted values of serum sodium concentration by the serum sodium concentration prediction model MO for 133 observation points. As shown in FIG. 7, the observation points are generally distributed near a straight line that indicates perfect prediction, and it can be said that very high prediction accuracy has been achieved. Note that the results shown in FIG. 7 are obtained by using linear support vector regression as a learning model and performing 10-fold cross validation as a verification method.
  • FIG. 8 is an explanatory diagram showing the prediction accuracy of the serum sodium concentration prediction model MO when the learning model and prediction factors are variously changed.
  • FIG. 8 shows the values of two evaluation indices (RMSE and R 2 ) representing the prediction accuracy of the serum sodium concentration prediction model MO for each combination of the learning model and the prediction factor.
  • Three learning model options are set: (1) linear regression, (2) linear support vector regression, and (3) bagging tree.
  • eight types of combinations selected from the seven predictor factor items described above are set as predictor options.
  • seven items of predictive factors are indicated by symbols (see FIG. 1).
  • items with black circles are items that were adopted as predictors, and items without black circles are items that were not adopted as predictors.
  • each combination of learning model and predictor showed generally high prediction accuracy.
  • the prediction accuracy was particularly high when linear regression and linear support vector regression were adopted.
  • the predictors as in the above embodiment, the prediction accuracy is highest when all seven items are adopted as predictors, but even if one or two of the seven items are omitted, , the decrease in prediction accuracy was minimal. Therefore, it can be said that even if the composition of the predictive factors differs somewhat due to differences in the equipment and operation of each medical institution, the serum sodium concentration can be predicted with high accuracy by using the serum sodium concentration prediction model MO.
  • the terminal device 100 of the present embodiment is an information processing device for predicting serum sodium concentration at a future time (t+ ⁇ t), and includes a predictive factor acquisition unit 114 and a prediction execution unit 115.
  • the predictive factor acquisition unit 114 acquires the measured value of the serum sodium concentration SSC(t) at the reference time t and the measured value of the serum sodium concentration SSC(t) at the reference time t and the future period T(0), which is the period from the reference time t to the future time (t+ ⁇ t), for the person to be predicted.
  • T(0) the future period from the reference time t to the future time (t+ ⁇ t)
  • the prediction execution unit 115 uses a serum sodium concentration prediction model MO generated by machine learning using training data TD in which a prediction factor is associated with a measured value of serum sodium concentration SSC (t+ ⁇ t) at a future time (t+ ⁇ t). , the serum sodium concentration SSC (t+ ⁇ t) of the subject at a future time (t+ ⁇ t) is predicted by inputting the predictive factors obtained for the subject.
  • the measured value of the serum sodium concentration SSC(t) at the reference time t and the index value representing the sodium intake in the future period T(0) for the person to be predicted By inputting the predictive factors including , into the serum sodium concentration prediction model MO, it is possible to accurately predict the subject's serum sodium concentration SSC (t+ ⁇ t) at future time (t+ ⁇ t). Therefore, according to the terminal device 100 of the present embodiment, it is possible to accurately predict the serum sodium concentration SSC (t+ ⁇ t) of a subject at a future time (t+ ⁇ t) by simply acquiring the above-mentioned predictive factors for the subject of prediction. Can be done.
  • the terminal device 100 of this embodiment further includes a prediction result output unit 116 that outputs a prediction result of the subject's serum sodium concentration. Therefore, according to the terminal device 100 of this embodiment, the user of the device can grasp the predicted value of the subject's serum sodium concentration SSC (t+ ⁇ t) at a future time (t+ ⁇ t).
  • the terminal device 100 of the present embodiment further includes a training data acquisition unit 112 that acquires training data TD, and a model acquisition unit 113 that creates a serum sodium concentration prediction model MO by machine learning using the training data TD. Be prepared. Therefore, according to the terminal device 100 of the present embodiment, the serum sodium concentration prediction model MO can be obtained without using any other device, and the serum sodium concentration of the subject at future time (t+ ⁇ t) can be calculated using this model. A prediction of SSC(t+ ⁇ t) can be performed.
  • the terminal device 100 of the present embodiment further includes updated training data that includes data in which predictive factors for the subject are associated with measured values of serum sodium concentration SSC (t+ ⁇ t) at future time (t+ ⁇ t).
  • a model updating unit 117 is provided that updates the serum sodium concentration prediction model MO by machine learning using TD. Therefore, according to the terminal device 100 of the present embodiment, the serum sodium concentration prediction model MO can be a model that is more suitable for the characteristics of the subject, and the accuracy of predicting the serum sodium concentration can be improved.
  • the predictive factor acquisition unit 114 acquires information specifying the type of ingestion (infusion fluid and/or drinking water) ingested by the subject, and the sodium content in each intake.
  • An index value representing the amount of sodium intake in the future period T(0) is obtained by referring to information indicating the amount (infusion information ID). Therefore, according to the terminal device 100 of the present embodiment, it is possible to realize the acquisition of the index value representing the sodium intake in the future period T(0) simply by the user specifying the type of ingestion (infusion). The prediction of serum sodium concentration can be performed more efficiently.
  • the predictive factors further include an index value representing the amount of water intake in the future period T(0) and a past period that is the period from the past time (t- ⁇ t) to the reference time t. and an index value representing the amount of water discharged during period T(-1). Therefore, according to the terminal device 100 of this embodiment, the prediction accuracy of serum sodium concentration can be improved.
  • the subject of prediction is a patient with hyponatremia. Therefore, according to the terminal device 100 of this embodiment, it is possible to predict the serum sodium concentration when a predetermined treatment is given to a hyponatremic patient.
  • the terminal device 100 of the present embodiment further predicts the subject's serum sodium concentration SSC(t+ ⁇ t) at a future time (t+ ⁇ t) from the measured value of the subject's serum sodium concentration SSC(t) at the reference time t.
  • the apparatus includes a notification section 118 that notifies you in a predetermined manner when the rate of increase in concentration to this value is faster than a preset threshold. Therefore, according to the terminal device 100 of the present embodiment, in treating a patient with hyponatremia, it is possible to suppress the rate of rise in serum sodium concentration from becoming excessively fast, and for example, to avoid the occurrence of ODS. I can do it.
  • the configuration of the terminal device 100 in the above embodiment is just an example, and can be modified in various ways. Further, the contents of the serum sodium concentration prediction model acquisition process and the serum sodium concentration prediction process in the above embodiment are merely examples, and can be modified in various ways.
  • the terminal device 100 obtains the serum sodium concentration prediction model MO by creating the serum sodium concentration prediction model MO.
  • the serum sodium concentration prediction model MO generated by the internal server device 200 or a device on an external network may be acquired. In this case, the terminal device 100 does not need to have the training data acquisition unit 112.
  • the serum sodium concentration prediction model MO is updated (S330 in FIG. 6), but the serum sodium concentration prediction model MO does not need to be updated. In this case, it is not necessary for the terminal device 100 to have the model updating section 117.
  • the notification process (S350 in FIG. 6) is executed, but the notification process does not need to be executed. In this case, the terminal device 100 does not need to have the notification section 118.
  • the predictive factors (FIG. 1) used to create the serum sodium concentration prediction model MO in the above embodiment are merely examples, and can be modified in various ways.
  • the predictive factors are the measured value of the serum potassium concentration SPC(t) at the reference time t, the measured value of the serum chloride concentration SCC(t) at the reference time t, and the potassium amount IPC contained in the infusion.
  • the predictor may not include at least one of these.
  • the urine volume UV is used as an index value representing the amount of water excreted in the past period T (-1), but instead of or in addition to this, other index values may be used. (For example, body weight change, plasma osmolarity, blood sugar level, etc.) may be used.
  • the infusion dose IV and/or the sodium amount ISC contained in the infusion in the future period T(0) is used as an index value representing the sodium intake in the future period T(0).
  • other index values for example, the intake amount of something ingested orally such as food and/or the amount of sodium contained in the object, etc. may be used. .
  • the terminal device 100 used by the medical worker P1 includes the serum sodium concentration prediction processing section 111 including the prediction factor acquisition section 114 and the prediction execution section 115.
  • At least some of the functions may exist not in the terminal device 100 but in another device (for example, the server device 200 in the in-hospital system 10 or a device on an external network).
  • the serum sodium concentration prediction processing section 111 may be incorporated as a function of the electronic medical record system.
  • the technology disclosed in this specification can be used for, for example, doctor-to-doctor remote medical consultation.
  • the terminal device 100 and/or the other device are an example of an information processing device (or information processing system) in the scope of the claims.
  • the device may predict other electrolyte concentrations (for example, potassium concentration) in the body and output them together with the predicted value of serum sodium concentration.
  • the above embodiment exemplifies information processing for predicting serum sodium concentration when a predetermined treatment is given to a patient with hyponatremia.
  • the present invention is not limited to, but can be similarly applied to prediction of serum sodium concentration in other situations.
  • the technology disclosed herein can be similarly applied to predicting the serum sodium concentration at a future time when oral supplementation is performed in an athletic person such as a long-distance runner. It is.
  • a part of the configuration realized by hardware may be replaced with software, or conversely, a part of the configuration realized by software may be replaced by hardware.
  • Terminal device 110 Control unit 111: Serum sodium concentration prediction processing unit 112: Training data acquisition unit 113: Model acquisition unit 114: Predictor acquisition unit 115: Prediction execution unit 116: Prediction result output unit 117: Model update section 118: Notification section 120: Storage section 130: Display section 140: Operation input section 150: Interface section 190: Bus 200: Server device

Landscapes

  • Health & Medical Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Medical Informatics (AREA)
  • Public Health (AREA)
  • General Health & Medical Sciences (AREA)
  • Epidemiology (AREA)
  • Primary Health Care (AREA)
  • Computing Systems (AREA)
  • Biomedical Technology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Pathology (AREA)
  • Business, Economics & Management (AREA)
  • Development Economics (AREA)
  • General Business, Economics & Management (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Economics (AREA)
  • Bioethics (AREA)
  • Accounting & Taxation (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Molecular Biology (AREA)
  • Surgery (AREA)
  • Animal Behavior & Ethology (AREA)
  • Veterinary Medicine (AREA)
  • Investigating Or Analysing Biological Materials (AREA)

Abstract

An information processing device for predicting serum sodium concentration at a future time comprises a predictor acquisition unit and a prediction execution unit. The predictor acquisition unit acquires, for a prediction subject, predictors including a measured value of serum sodium concentration at a reference time, and an index value representing sodium intake at a future period, which is a period from the reference time to a future time. The prediction execution unit predicts the subject's serum sodium concentration at the future time by inputting the predictors acquired for the subject into a serum sodium concentration prediction model generated by machine learning using training data associating predictors and future measurement values of serum sodium concentration.

Description

情報処理装置、情報処理方法、および、コンピュータプログラムInformation processing device, information processing method, and computer program
　本明細書に開示される技術は、将来時刻における血清ナトリウム濃度を予測するための情報処理装置等に関する。 The technology disclosed herein relates to an information processing device and the like for predicting serum sodium concentration at a future time.
　低ナトリウム血症は、血清ナトリウム濃度が所定値（例えば、１３５ｍＥｑ／Ｌ）より低くなる体内電解質異常であり、例えば、嘔気、頭痛、ふらつき等の症状を伴う。重度の低ナトリウム血症（例えば、血清ナトリウム濃度が１２５ｍＥｑ／Ｌより低い状態）は、意識障害、痙攣、昏睡等の重篤な中枢神経症状の原因となり得る。 Hyponatremia is an electrolyte abnormality in the body in which the serum sodium concentration is lower than a predetermined value (eg, 135 mEq/L), and is accompanied by symptoms such as nausea, headache, and light-headedness. Severe hyponatremia (eg, serum sodium concentration below 125 mEq/L) can cause severe central nervous system symptoms such as impaired consciousness, convulsions, and coma.
　一般に、低ナトリウム血症は、（１）細胞外液量の減少を伴うもの、（２）細胞外液量がほぼ正常であるもの、（３）細胞外液量の増加を伴うもの、の３つに大別される。細胞外液減量の減少を伴う低ナトリウム血症に対する治療としては、輸液（例えば、生理食塩水、高張（３％）食塩水、３号液、ブドウ糖液等）の投与による細胞外液の補充が行われる。一方、細胞外液量がほぼ正常もしくは細胞外液量の増加を伴う低ナトリウム血症に対する治療としては、水制限やバソプレシンＶ２受容体拮抗薬による体内に貯留される自由水の排泄促進が行われる。このように両者に対する治療法は正反対となる。そのため、個々の症例において、正確に病態を把握して低ナトリウム血症の鑑別を行い、各病態に応じた適切な治療を行うことが重要である。しかしながら、細胞外液量を正確に評価することは極めて難しいため、治療開始前に細胞外液量に基づく病態の把握や低ナトリウム血症の鑑別を行うことは難しい。また、複数の病態が併存している症例や、病初期から治療中にかけて病態が変化する症例も存在する。 In general, hyponatremia can be categorized into three types: (1) a decrease in extracellular fluid volume, (2) an almost normal extracellular fluid volume, and (3) an increase in extracellular fluid volume. It is broadly divided into Hyponatremia associated with decreased extracellular fluid loss can be treated by replenishing extracellular fluid by administering intravenous fluids (e.g., physiological saline, hypertonic (3%) saline, No. 3 solution, glucose solution, etc.). It will be done. On the other hand, treatments for hyponatremia with a nearly normal or increased extracellular fluid volume include water restriction and promotion of excretion of free water stored in the body using vasopressin V2 receptor antagonists. . Thus, the treatments for both are diametrically opposed. Therefore, in each individual case, it is important to accurately understand the pathological condition, differentiate hyponatremia, and provide appropriate treatment according to each pathological condition. However, since it is extremely difficult to accurately evaluate the extracellular fluid volume, it is difficult to understand the pathological condition or differentiate hyponatremia based on the extracellular fluid volume before starting treatment. Furthermore, there are cases in which multiple pathological conditions coexist, and cases in which the pathological condition changes from the initial stage of the disease to the course of treatment.
　また、低ナトリウム血症の治療において、血清ナトリウム濃度を過度に速い速度で（例えば、８～１０ｍＥｑ／Ｌ／日以上の速度で）上昇させると、浸透圧性脱髄症候群（Ｏｓｍｏｔｉｃ　Ｄｅｍｙｅｌｉｎａｔｉｏｎ　Ｓｙｎｄｒｏｍｅ：ＯＤＳ）を引き起こすおそれがある。ＯＤＳは、構音障害、嚥下障害、四肢麻痺、意識障害等の重篤な症状を呈し、致死的となることも少なくない上に、有効性が確立された治療法が存在しない。そのため、低ナトリウム血症の治療においては、ＯＤＳを予防するために、血清ナトリウム濃度の急激な上昇を避けることが極めて重要である。 In addition, in the treatment of hyponatremia, raising the serum sodium concentration at an excessively rapid rate (for example, at a rate of 8 to 10 mEq/L/day or more) may cause Osmotic Demyelination Syndrome (ODS). may cause. ODS exhibits serious symptoms such as dysarthria, dysphagia, quadriplegia, and impaired consciousness, and is often fatal, and there is no proven effective treatment. Therefore, in the treatment of hyponatremia, it is extremely important to avoid rapid increases in serum sodium concentration in order to prevent ODS.
　しかしながら、上述したように、実臨床においては、治療開始前に正確に病態を把握して低ナトリウム血症の鑑別を行うことが難しく、また、治療中に病態が変化することもある。そのため、治療中の患者を対象として、頻回に血清ナトリウム濃度をモニタリングしたとしても、血清ナトリウム濃度の急激な上昇が発生するおそれがある。また、夜間に頻回の血液検査を行うことが困難な医療機関も存在する。 However, as mentioned above, in actual clinical practice, it is difficult to accurately grasp the pathological condition and differentiate hyponatremia before starting treatment, and the pathological condition may change during treatment. Therefore, even if serum sodium concentration is frequently monitored in patients undergoing treatment, there is a risk that a rapid increase in serum sodium concentration may occur. Furthermore, there are some medical institutions where it is difficult to conduct frequent blood tests at night.
　従来、電子カルテの情報から種々の医療イベントを予測するためのシステムが提案されている（例えば、特許文献１参照））。 Conventionally, systems for predicting various medical events from information in electronic medical records have been proposed (see, for example, Patent Document 1).
特表２０２０－５２９０５７号公報Special Publication No. 2020-529057
　しかしながら、上記従来のシステムは、将来時刻における血清ナトリウム濃度を予測するものではない。低ナトリウム血症の治療を適切に行うために、将来時刻における血清ナトリウム濃度を精度良く予測する技術が望まれている。 However, the conventional system described above does not predict serum sodium concentration at a future time. In order to appropriately treat hyponatremia, a technology that accurately predicts serum sodium concentration at a future time is desired.
　なお、このような課題は、治療中の低ナトリウム血症の患者を対象とした将来時刻における血清ナトリウム濃度の予測に限られるものではない。例えば長距離ランナーといった運動をしている人を対象として、経口による補給を行ったときの将来時刻における血清ナトリウム濃度を予測する際にも同様の課題がある。すなわち、人全般を対象として、将来時刻における血清ナトリウム濃度を精度良く予測する技術が知られていないという課題がある。 Note that such a problem is not limited to predicting serum sodium concentration at a future time for hyponatremic patients undergoing treatment. A similar problem exists when predicting serum sodium levels at future times when oral supplementation is performed in physically active people, such as long-distance runners. That is, there is a problem in that there is no known technique for accurately predicting serum sodium concentration at a future time for all people.
　本明細書では、上述した課題を解決することが可能な技術を開示する。 This specification discloses a technique that can solve the above-mentioned problems.
　本明細書に開示される技術は、例えば、以下の形態として実現することが可能である。 The technology disclosed in this specification can be realized, for example, in the following form.
（１）本明細書に開示される情報処理装置は、将来時刻における血清ナトリウム濃度を予測するための装置であって、予測因子取得部と、予測実行部とを備える。予測因子取得部は、前記予測の対象者について、基準時刻における血清ナトリウム濃度の測定値と、前記基準時刻から前記将来時刻までの期間である将来期間におけるナトリウム摂取量を表す指標値と、を含む予測因子を取得する。予測実行部は、前記予測因子と前記将来時刻における血清ナトリウム濃度の測定値とが対応付けられた訓練データを用いた機械学習により生成された血清ナトリウム濃度予測モデルに、前記対象者について取得された前記予測因子を入力することにより、前記将来時刻における前記対象者の血清ナトリウム濃度を予測する。 (1) The information processing device disclosed in this specification is a device for predicting serum sodium concentration at a future time, and includes a prediction factor acquisition unit and a prediction execution unit. The predictive factor acquisition unit includes, for the subject of the prediction, a measured value of serum sodium concentration at a reference time, and an index value representing sodium intake in a future period that is a period from the reference time to the future time. Get predictors. The prediction execution unit applies the serum sodium concentration prediction model generated by machine learning using training data in which the predictive factors and the measured value of serum sodium concentration at the future time are associated with the serum sodium concentration obtained for the subject. By inputting the predictive factor, the subject's serum sodium concentration at the future time is predicted.
　このように、本情報処理装置では、予測の対象者についての、基準時刻における血清ナトリウム濃度の測定値と、将来期間におけるナトリウム摂取量を表す指標値と、を含む予測因子を、血清ナトリウム濃度予測モデルに入力することにより、将来時刻における対象者の血清ナトリウム濃度を精度良く予測することができる。 In this way, this information processing device uses predictive factors that include the measured value of serum sodium concentration at the reference time and the index value representing sodium intake in the future period for the subject of prediction to predict serum sodium concentration. By inputting it into the model, it is possible to accurately predict the subject's serum sodium concentration at a future time.
（２）上記情報処理装置において、さらに、前記対象者の血清ナトリウム濃度の予測結果を出力する予測結果出力部を備える構成としてもよい。本構成を採用すれば、装置の使用者に、将来時刻における対象者の血清ナトリウム濃度の予測値を把握させることができる。 (2) The information processing apparatus may further include a prediction result output unit that outputs a prediction result of the subject's serum sodium concentration. By employing this configuration, the user of the device can be made aware of the predicted value of the subject's serum sodium concentration at a future time.
（３）上記情報処理装置において、さらに、前記訓練データを取得する訓練データ取得部と、前記訓練データを用いた前記機械学習によって前記血清ナトリウム濃度予測モデルを作成するモデル取得部と、を備える構成としてもよい。本構成を採用すれば、他の装置を用いずとも血清ナトリウム濃度予測モデルを取得することができ、該モデルを用いて将来時刻における対象者の血清ナトリウム濃度の予測を実行することができる。 (3) The information processing device further includes a training data acquisition unit that acquires the training data, and a model acquisition unit that creates the serum sodium concentration prediction model by the machine learning using the training data. You can also use it as If this configuration is adopted, a serum sodium concentration prediction model can be obtained without using any other device, and the serum sodium concentration of the subject at a future time can be predicted using this model.
（４）上記情報処理装置において、さらに、前記対象者についての前記予測因子と前記将来時刻における血清ナトリウム濃度の測定値とが対応付けられたデータを含む更新された前記訓練データを用いた前記機械学習によって、前記血清ナトリウム濃度予測モデルを更新するモデル更新部を備える構成としてもよい。本構成を採用すれば、血清ナトリウム濃度予測モデルを、対象者の特性により適合したモデルとすることができ、血清ナトリウム濃度の予測精度を向上させることができる。 (4) In the information processing device, the machine further uses the updated training data including data in which the predictive factor for the subject is associated with a measured value of serum sodium concentration at the future time. The configuration may include a model updating section that updates the serum sodium concentration prediction model through learning. If this configuration is adopted, the serum sodium concentration prediction model can be made into a model that is more suitable for the characteristics of the subject, and the prediction accuracy of the serum sodium concentration can be improved.
（５）上記情報処理装置において、前記予測因子取得部は、前記対象者が摂取する摂取物の種類を特定する情報を取得すると共に、各前記摂取物に含まれるナトリウム量を示す情報を参照して、前記将来期間におけるナトリウム摂取量を表す指標値を取得する構成としてもよい。本構成を採用すれば、使用者が摂取物の種類を指定するだけで、将来期間におけるナトリウム摂取量を表す指標値の取得を実現することができ、より効率的に血清ナトリウム濃度の予測を実行することができる。 (5) In the information processing device, the predictive factor acquisition unit acquires information specifying the type of ingestion that the subject ingests, and also refers to information indicating the amount of sodium contained in each of the ingestions. Then, an index value representing the amount of sodium intake in the future period may be acquired. By adopting this configuration, the user can obtain an index value representing sodium intake in the future period simply by specifying the type of food consumed, and more efficiently predict serum sodium concentration. can do.
（６）上記情報処理装置において、前記予測因子は、さらに、前記将来期間における水分摂取量を表す指標値と、過去時刻から前記基準時刻までの期間である過去期間における水分排出量を表す指標値と、を含む構成としてもよい。本構成を採用すれば、血清ナトリウム濃度の予測精度を向上させることができる。 (6) In the information processing device, the predictor further includes an index value representing the amount of water intake in the future period, and an index value representing the amount of water discharged in the past period, which is a period from the past time to the reference time. It is also possible to have a configuration including the following. If this configuration is adopted, the accuracy of predicting serum sodium concentration can be improved.
（７）上記情報処理装置において、前記対象者は、低ナトリウム血症の患者である構成としてもよい。本構成を採用すれば、低ナトリウム血症の患者に対して所定の治療を施したときの血清ナトリウム濃度の予測を実現することができる。 (7) In the information processing device, the subject may be a hyponatremic patient. By employing this configuration, it is possible to predict the serum sodium concentration when a predetermined treatment is given to a hyponatremic patient.
（８）上記情報処理装置において、さらに、前記基準時刻における前記対象者の血清ナトリウム濃度の測定値から、前記将来時刻における前記対象者の血清ナトリウム濃度の予測値への濃度上昇速度が予め設定された閾値より速い場合に、所定の方法で報知する報知部を備える構成としてもよい。本構成を採用すれば、低ナトリウム血症の患者の治療において、血清ナトリウム濃度の上昇速度が過度に速くなることを抑制することができ、例えばＯＤＳの発生を回避することができる。 (8) In the information processing device, a rate of increase in concentration from a measured value of the serum sodium concentration of the subject at the reference time to a predicted value of the serum sodium concentration of the subject at the future time is further set in advance. The configuration may include a notification section that provides notification using a predetermined method when the speed is faster than a threshold value. By employing this configuration, it is possible to suppress the rate of increase in serum sodium concentration from becoming excessively fast in treating patients with hyponatremia, and for example, it is possible to avoid the occurrence of ODS.
　なお、本明細書に開示される技術は、種々の形態で実現することが可能であり、例えば、情報処理装置、情報処理方法、それらの方法を実現するコンピュータプログラム、そのコンピュータプログラムを記録した一時的でない記録媒体等の形態で実現することができる。 Note that the technology disclosed in this specification can be realized in various forms, such as an information processing device, an information processing method, a computer program that implements these methods, and a temporary computer program that records the computer program. It can be realized in the form of a non-standard recording medium or the like.
血清ナトリウム濃度予測モデルＭＯを概念的に示す説明図Explanatory diagram conceptually showing the serum sodium concentration prediction model MO 院内システム１０の構成を示す説明図Explanatory diagram showing the configuration of the hospital system 10 端末装置１００の構成を概略的に示すブロック図A block diagram schematically showing the configuration of a terminal device 100 輸液情報ＩＤの一例を示す説明図Explanatory diagram showing an example of infusion information ID 血清ナトリウム濃度予測モデル取得処理を示すフローチャートFlowchart showing the serum sodium concentration prediction model acquisition process 血清ナトリウム濃度予測処理を示すフローチャートFlowchart showing serum sodium concentration prediction process 本実施例の血清ナトリウム濃度予測モデルＭＯの予測精度を示す説明図Explanatory diagram showing the prediction accuracy of the serum sodium concentration prediction model MO of this example 学習モデルおよび予測因子を種々変更した場合における血清ナトリウム濃度予測モデルＭＯの予測精度を示す説明図Explanatory diagram showing the prediction accuracy of the serum sodium concentration prediction model MO when the learning model and predictive factors are variously changed
Ａ．実施形態：
Ａ－１．血清ナトリウム濃度予測モデルＭＯの概要：
　はじめに、本実施形態における血清ナトリウム濃度予測モデルＭＯの概要を説明する。図１は、血清ナトリウム濃度予測モデルＭＯを概念的に示す説明図である。 A. Embodiment:
A-1. Overview of serum sodium concentration prediction model MO:
First, an overview of the serum sodium concentration prediction model MO in this embodiment will be explained. FIG. 1 is an explanatory diagram conceptually showing a serum sodium concentration prediction model MO.
　血清ナトリウム濃度予測モデルＭＯは、将来時刻における血清ナトリウム濃度を予測するために用いられる学習済みモデルである。図１に示すように、血清ナトリウム濃度予測モデルＭＯは、７つの予測因子を入力とし、将来時刻（ｔ＋Δｔ）における血清ナトリウム濃度ＳＳＣ（ｔ＋Δｔ）（ｍＥｑ／Ｌ）を出力（応答）とする機械学習モデルである。本実施形態では、入力としての予測因子は、以下の７項目を含む。
・基準時刻ｔにおける血清ナトリウム濃度ＳＳＣ（ｔ）の測定値（ｍＥｑ／Ｌ）
・基準時刻ｔにおける血清カリウム濃度ＳＰＣ（ｔ）の測定値（ｍＥｑ／Ｌ）
・基準時刻ｔにおける血清クロール濃度ＳＣＣ（ｔ）の測定値（ｍＥｑ／Ｌ）
・基準時刻ｔから将来時刻（ｔ＋Δｔ）までの将来期間Ｔ（０）における輸液投与量ＩＶ（ｍｌ）
・輸液に含まれるナトリウム量ＩＳＣ（ｍＥｑ／Ｌ）
・輸液に含まれるカリウム量ＩＰＣ（ｍＥｑ／Ｌ）
・過去時刻（ｔ－Δｔ）から基準時刻ｔまでの過去期間Ｔ（－１）における尿量ＵＶ（ｍｌ） The serum sodium concentration prediction model MO is a trained model used to predict the serum sodium concentration at a future time. As shown in Figure 1, the serum sodium concentration prediction model MO is a machine learning method that uses seven predictors as input and outputs (response) the serum sodium concentration SSC (t+Δt) (mEq/L) at future time (t+Δt). It's a model. In this embodiment, the predictive factors as input include the following seven items.
・Measurement value of serum sodium concentration SSC (t) at reference time t (mEq/L)
・Measurement value of serum potassium concentration SPC (t) at reference time t (mEq/L)
・Measurement value of serum chloride concentration SCC (t) at reference time t (mEq/L)
- Infusion dose IV (ml) in future period T (0) from reference time t to future time (t + Δt)
・Sodium content ISC (mEq/L) contained in the infusion
・Amount of potassium contained in infusion IPC (mEq/L)
・Urine volume UV (ml) in the past period T (-1) from past time (t-Δt) to reference time t
　なお、基準時刻ｔは、将来時刻（ｔ＋Δｔ）における血清ナトリウム濃度ＳＳＣ（ｔ＋Δｔ）の予測の際の基準となる時刻であり、血清ナトリウム濃度ＳＳＣ（ｔ）の測定値が既知の時刻である。基準時刻ｔは、任意の時刻を取り得るが、例えば血清ナトリウム濃度予測モデルＭＯを用いて血清ナトリウム濃度の予測を行おうとしている時刻である。また、基準時刻ｔと将来時刻（ｔ＋Δｔ）との時間間隔、および、過去時刻（ｔ－Δｔ）と基準時刻ｔとの時間間隔であるΔｔは、任意の値を取り得るが、本実施形態では固定値の６時間に設定される。Δｔは、可変値であってもよい。また、将来時刻（ｔ＋Δｔ）は、基準時刻ｔから見た将来の時刻であればよく、必ずしも実際の「将来の」時刻には限られない。例えば、現在からΔｔだけ遡った時刻は、現在から（Δｔ×２）だけ遡った時刻を基準時刻ｔとすれば、該基準時刻ｔから見た「将来時刻（ｔ＋Δｔ）」に該当する。 Note that the reference time t is a time that serves as a reference for predicting the serum sodium concentration SSC(t+Δt) at a future time (t+Δt), and is a time at which the measured value of the serum sodium concentration SSC(t) is known. The reference time t can be any arbitrary time, but is, for example, the time at which the serum sodium concentration is predicted using the serum sodium concentration prediction model MO. Further, the time interval between the reference time t and the future time (t+Δt), and the time interval between the past time (t−Δt) and the reference time t, Δt, can take any value, but in this embodiment, It is set to a fixed value of 6 hours. Δt may be a variable value. Further, the future time (t+Δt) may be any future time seen from the reference time t, and is not necessarily limited to an actual "future" time. For example, if the reference time t is a time that is (Δt×2) from the present, a time that is Δt back from the present corresponds to “future time (t+Δt)” as viewed from the reference time t.
　また、上記７つの予測因子のうち、「基準時刻ｔから将来時刻（ｔ＋Δｔ）までの将来期間Ｔ（０）における輸液投与量ＩＶ」および／または「輸液に含まれるナトリウム量ＩＳＣ」は、将来期間Ｔ（０）におけるナトリウム摂取量を表す指標値であると言える。また、「将来期間Ｔ（０）における輸液投与量ＩＶ」は、将来期間Ｔ（０）における水分摂取量を表す指標値であると言える。また、「過去時刻（ｔ－Δｔ）から基準時刻ｔまでの過去期間Ｔ（－１）における尿量ＵＶ」は、過去期間Ｔ（－１）における水分排出量を表す指標値であると言える。そのため、本実施形態では、血清ナトリウム濃度予測モデルＭＯの入力としての予測因子は、基準時刻ｔにおける血清ナトリウム濃度の測定値と、基準時刻ｔから将来時刻（ｔ＋Δｔ）までの将来期間Ｔ（０）におけるナトリウム摂取量を表す指標値と、将来期間Ｔ（０）における水分摂取量を表す指標値と、過去時刻（ｔ－Δｔ）から基準時刻ｔまでの過去期間Ｔ（－１）における水分排出量を表す指標値と、を含む。 In addition, among the above seven predictors, "infusion dose IV in future period T (0) from reference time t to future time (t + Δt)" and/or "sodium amount ISC contained in infusion" It can be said that this is an index value representing the sodium intake at T(0). Furthermore, it can be said that the "infusion dose IV in the future period T(0)" is an index value representing the water intake amount in the future period T(0). Furthermore, it can be said that "the urine volume UV in the past period T(-1) from the past time (t-Δt) to the reference time t" is an index value representing the amount of water excreted in the past period T(-1). Therefore, in this embodiment, the predictive factors as inputs of the serum sodium concentration prediction model MO are the measured value of the serum sodium concentration at the reference time t and the future period T(0) from the reference time t to the future time (t+Δt). An index value representing the sodium intake in the future period T(0), an index value representing the water intake in the future period T(0), and a water excretion amount in the past period T(-1) from the past time (t-Δt) to the reference time t. and an index value representing.
Ａ－２．院内システム１０の構成：
　次に、院内システム１０の構成について説明する。図２は、院内システム１０の構成を示す説明図である。院内システム１０は、病院内に導入された情報システムである。本実施形態の院内システム１０では、低ナトリウム血症の患者を対象として、上述した血清ナトリウム濃度予測モデルＭＯを用いて、患者に所定の治療を施したときの将来時刻における血清ナトリウム濃度の予測が実行される。 A-2. Configuration of hospital system 10:
Next, the configuration of the in-hospital system 10 will be explained. FIG. 2 is an explanatory diagram showing the configuration of the in-hospital system 10. The in-hospital system 10 is an information system introduced within a hospital. The in-hospital system 10 of this embodiment uses the above-described serum sodium concentration prediction model MO to predict the serum sodium concentration at a future time when a predetermined treatment is given to the patient. executed.
　図２に示すように、院内システム１０は、医師や看護師等の医療従事者Ｐ１が使用する端末装置１００と、院内に設置されたサーバ装置２００とを備える。院内システム１０を構成する各装置は、通信ネットワークＮＥＴを介して互いに通信可能に接続されている。 As shown in FIG. 2, the in-hospital system 10 includes a terminal device 100 used by a medical worker P1 such as a doctor or nurse, and a server device 200 installed within the hospital. The devices constituting the in-hospital system 10 are communicably connected to each other via the communication network NET.
　端末装置１００は、例えば、ＰＣやタブレット型端末、スマートフォン等である。図３は、端末装置１００の構成を概略的に示すブロック図である。端末装置１００は、制御部１１０と、記憶部１２０と、表示部１３０と、操作入力部１４０と、インターフェース部１５０とを備える。これらの各部は、バス１９０を介して互いに通信可能に接続されている。端末装置１００は、特許請求の範囲における情報処理装置の一例である。 The terminal device 100 is, for example, a PC, a tablet terminal, a smartphone, or the like. FIG. 3 is a block diagram schematically showing the configuration of the terminal device 100. The terminal device 100 includes a control section 110, a storage section 120, a display section 130, an operation input section 140, and an interface section 150. These units are communicably connected to each other via a bus 190. The terminal device 100 is an example of an information processing device within the scope of the claims.
　端末装置１００の表示部１３０は、例えば液晶ディスプレイ等により構成され、各種の画像や情報を表示する。操作入力部１４０は、例えばキーボードやマウス、ボタン、マイク等により構成され、医療従事者Ｐ１の操作や指示を受け付ける。なお、表示部１３０がタッチパネルを備えることにより、操作入力部１４０として機能するとしてもよい。インターフェース部１５０は、例えばＬＡＮインターフェースやＵＳＢインターフェース等により構成され、有線または無線により他の装置との通信を行う。 The display unit 130 of the terminal device 100 is composed of, for example, a liquid crystal display, and displays various images and information. The operation input unit 140 includes, for example, a keyboard, a mouse, buttons, a microphone, etc., and receives operations and instructions from the medical worker P1. Note that the display unit 130 may function as the operation input unit 140 by including a touch panel. The interface section 150 is configured with, for example, a LAN interface, a USB interface, etc., and communicates with other devices by wire or wirelessly.
　端末装置１００の記憶部１２０は、例えばＲＯＭやＲＡＭ、ハードディスクドライブ（ＨＤＤ）、ソリッドステートドライブ（ＳＳＤ）等により構成され、各種のプログラムやデータを記憶したり、各種のプログラムを実行する際の作業領域やデータの一時的な記憶領域として利用されたりする。例えば、記憶部１２０には、後述する種々の処理を実行するためのコンピュータプログラムである血清ナトリウム濃度予測プログラムＣＰが格納されている。血清ナトリウム濃度予測プログラムＣＰは、例えば、ＣＤ－ＲＯＭやＤＶＤ－ＲＯＭ、ＵＳＢメモリ等のコンピュータ読み取り可能な記録媒体（不図示）に格納された状態で提供され、あるいは、インターフェース部１５０を介して外部装置（例えば、クラウド上のサーバや他の端末装置）から取得可能な状態で提供され、端末装置１００上で動作可能な状態で記憶部１２０に格納される。 The storage unit 120 of the terminal device 100 is composed of, for example, ROM, RAM, hard disk drive (HDD), solid state drive (SSD), etc., and stores various programs and data, and performs work when executing various programs. It is used as a temporary storage area for space and data. For example, the storage unit 120 stores a serum sodium concentration prediction program CP, which is a computer program for executing various processes described below. The serum sodium concentration prediction program CP is provided, for example, in a state stored in a computer-readable recording medium (not shown) such as a CD-ROM, DVD-ROM, or USB memory, or is provided externally via the interface section 150. It is provided in a state that can be obtained from a device (for example, a server on a cloud or another terminal device), and is stored in the storage unit 120 in a state that can be operated on the terminal device 100.
　また、端末装置１００の記憶部１２０には、後述する種々の処理において、訓練データＴＤと、血清ナトリウム濃度予測モデルＭＯと、予測因子データＰＤと、予測結果データＲＤとが格納される。これらのデータやモデルについては、後述の種々の処理の説明に合わせて説明する。 Furthermore, the storage unit 120 of the terminal device 100 stores training data TD, serum sodium concentration prediction model MO, predictive factor data PD, and prediction result data RD in various processes described below. These data and models will be explained in conjunction with explanations of various processes later.
　端末装置１００の制御部１１０は、例えばＣＰＵ等により構成され、記憶部１２０から読み出したコンピュータプログラムを実行することにより、端末装置１００の動作を制御する。例えば、制御部１１０は、記憶部１２０から血清ナトリウム濃度予測プログラムＣＰを読み出して実行することにより、後述の種々の処理を実行する血清ナトリウム濃度予測処理部１１１として機能する。血清ナトリウム濃度予測処理部１１１は、訓練データ取得部１１２と、モデル取得部１１３と、予測因子取得部１１４と、予測実行部１１５と、予測結果出力部１１６と、モデル更新部１１７と、報知部１１８とを含む。これら各部の機能については、後述の種々の処理の説明に合わせて説明する。 The control unit 110 of the terminal device 100 is configured by, for example, a CPU, and controls the operation of the terminal device 100 by executing a computer program read from the storage unit 120. For example, the control unit 110 reads the serum sodium concentration prediction program CP from the storage unit 120 and executes it, thereby functioning as a serum sodium concentration prediction processing unit 111 that executes various processes described below. The serum sodium concentration prediction processing section 111 includes a training data acquisition section 112, a model acquisition section 113, a predictive factor acquisition section 114, a prediction execution section 115, a prediction result output section 116, a model update section 117, and a notification section. 118. The functions of these units will be explained in conjunction with explanations of various processes later.
　サーバ装置２００（図２）は、院内システム１０において情報提示機能や情報処理機能を提供する装置である。本実施形態では、サーバ装置２００は、輸液情報ＩＤを格納している。図４は、輸液情報ＩＤの一例を示す説明図である。輸液情報ＩＤは、低ナトリウム血症の患者に投与する輸液の種類（製剤）と、各種類の輸液に含まれるナトリウムおよびカリウムの量（ｍＥｑ／Ｌ）とを対応付ける情報である。輸液情報ＩＤを参照することにより、各種類の輸液に含まれるナトリウムおよびカリウムの量を特定することができる。 The server device 200 (FIG. 2) is a device that provides an information presentation function and an information processing function in the in-hospital system 10. In this embodiment, the server device 200 stores infusion information ID. FIG. 4 is an explanatory diagram showing an example of infusion information ID. The infusion information ID is information that associates the type (preparation) of an infusion to be administered to a hyponatremic patient with the amount of sodium and potassium (mEq/L) contained in each type of infusion. By referring to the infusion information ID, the amounts of sodium and potassium contained in each type of infusion can be specified.
Ａ－３．血清ナトリウム濃度予測モデル取得処理：
　次に、本実施形態の端末装置１００により実行される血清ナトリウム濃度予測モデル取得処理について説明する。図５は、血清ナトリウム濃度予測モデル取得処理を示すフローチャートである。血清ナトリウム濃度予測モデル取得処理は、上述した血清ナトリウム濃度予測モデルＭＯを取得するための処理である。本実施形態では、端末装置１００が、自ら所定の機械学習によって血清ナトリウム濃度予測モデルＭＯを作成することにより、血清ナトリウム濃度予測モデルＭＯを取得する。血清ナトリウム濃度予測モデル取得処理は、医療従事者Ｐ１が端末装置１００の操作入力部１４０を操作して開始指示を入力したことに応じて開始される。 A-3. Serum sodium concentration prediction model acquisition process:
Next, a serum sodium concentration prediction model acquisition process executed by the terminal device 100 of this embodiment will be described. FIG. 5 is a flowchart showing the serum sodium concentration prediction model acquisition process. The serum sodium concentration prediction model acquisition process is a process for acquiring the serum sodium concentration prediction model MO mentioned above. In this embodiment, the terminal device 100 acquires the serum sodium concentration prediction model MO by creating the serum sodium concentration prediction model MO by itself using predetermined machine learning. The serum sodium concentration prediction model acquisition process is started in response to the medical worker P1 operating the operation input unit 140 of the terminal device 100 to input a start instruction.
　はじめに、端末装置１００の訓練データ取得部１１２（図３）が、訓練データＴＤを取得する（Ｓ１１０）。訓練データＴＤは、図１に示す７つの予測因子と、将来時刻（ｔ＋Δｔ）における血清ナトリウム濃度ＳＳＣ（ｔ＋Δｔ）の測定値（実績値）とが対応付けられた複数のデータの集合である。訓練データＴＤは、インターフェース部１５０を介して取得され、記憶部１２０に格納される。 First, the training data acquisition unit 112 (FIG. 3) of the terminal device 100 acquires training data TD (S110). The training data TD is a set of multiple data in which the seven predictive factors shown in FIG. 1 are associated with the measured value (actual value) of the serum sodium concentration SSC (t+Δt) at a future time (t+Δt). Training data TD is acquired via the interface section 150 and stored in the storage section 120.
　次に、端末装置１００のモデル取得部１１３（図３）が、訓練データＴＤを用いた所定の機械学習（深層学習を含む）により、血清ナトリウム濃度予測モデルＭＯを作成する（Ｓ１２０）。モデル取得部１１３は、訓練データＴＤに含まれる７つの予測因子を説明変数とし、訓練データＴＤに含まれる将来時刻（ｔ＋Δｔ）における血清ナトリウム濃度ＳＳＣ（ｔ＋Δｔ）の測定値を目的変数とし、所定の評価指標（例えば、平均二乗平方根誤差（ＲＭＳＥ）、平均絶対誤差（ＭＡＥ）、決定係数（Ｒ２））を参照しつつ所定の学習アルゴリズムに基づき機械学習を実行することにより、血清ナトリウム濃度予測モデルＭＯを作成する。血清ナトリウム濃度予測モデルＭＯの作成には、公知の種々の学習アルゴリズムを利用可能であるが、例えば、サポートベクター回帰、線形回帰、バギング木、ニューラルネットワーク(ディープニューラルネットワークを含む)等を用いることができる。作成された血清ナトリウム濃度予測モデルＭＯは、端末装置１００の記憶部１２０に格納される。以上の工程により、血清ナトリウム濃度予測モデル取得処理が完了する。 Next, the model acquisition unit 113 (FIG. 3) of the terminal device 100 creates a serum sodium concentration prediction model MO by predetermined machine learning (including deep learning) using the training data TD (S120). The model acquisition unit 113 uses the seven predictive factors included in the training data TD as explanatory variables, uses the measured value of serum sodium concentration SSC (t+Δt) at a future time (t+Δt) included in the training data TD as an objective variable, and uses a predetermined A serum sodium concentration prediction model is created by performing machine learning based on a predetermined learning algorithm while referring to evaluation indicators (e.g. root mean square error (RMSE), mean absolute error (MAE), coefficient of determination (R 2 )). Create MO. Various known learning algorithms can be used to create the serum sodium concentration prediction model MO; for example, support vector regression, linear regression, bagging trees, neural networks (including deep neural networks), etc. can be used. can. The created serum sodium concentration prediction model MO is stored in the storage unit 120 of the terminal device 100. Through the above steps, the serum sodium concentration prediction model acquisition process is completed.
Ａ－４．血清ナトリウム濃度予測処理：
　次に、本実施形態の端末装置１００により実行される血清ナトリウム濃度予測処理について説明する。図６は、血清ナトリウム濃度予測処理を示すフローチャートである。血清ナトリウム濃度予測処理は、将来時刻における血清ナトリウム濃度を予測する処理である。より具体的には、血清ナトリウム濃度予測処理は、低ナトリウム血症の患者を対象として、患者に所定の治療を施したときの将来時刻（ｔ＋Δｔ）における血清ナトリウム濃度ＳＳＣ（ｔ＋Δｔ）を、血清ナトリウム濃度予測モデルＭＯを用いて予測する処理である。血清ナトリウム濃度予測処理は、医療従事者Ｐ１が端末装置１００の操作入力部１４０を操作して開始指示を入力したことに応じて開始される。 A-4. Serum sodium concentration prediction processing:
Next, the serum sodium concentration prediction process executed by the terminal device 100 of this embodiment will be described. FIG. 6 is a flowchart showing the serum sodium concentration prediction process. The serum sodium concentration prediction process is a process for predicting the serum sodium concentration at a future time. More specifically, the serum sodium concentration prediction process is based on the serum sodium concentration SSC (t+Δt) at a future time (t+Δt) when a predetermined treatment is given to the patient with hyponatremia. This is a process of predicting using the concentration prediction model MO. The serum sodium concentration prediction process is started in response to the medical worker P1 operating the operation input unit 140 of the terminal device 100 and inputting a start instruction.
　はじめに、端末装置１００の予測因子取得部１１４（図３）が、予測の対象者である低ナトリウム血症の患者について、上述した７つの予測因子（図１）を取得する（Ｓ３１０）。取得された予測因子は、予測因子データＰＤとして記憶部１２０に格納される。 First, the predictive factor acquisition unit 114 (FIG. 3) of the terminal device 100 acquires the seven predictive factors (FIG. 1) described above for a hyponatremic patient who is a prediction target (S310). The acquired predictive factors are stored in the storage unit 120 as predictive factor data PD.
　具体的には、７つの予測因子のうち、以下の３つの予測因子については、医療従事者Ｐ１が患者を対象として血液検査を行い、操作入力部１４０を介して検査結果（各測定値）を入力する。予測因子取得部１１４は、入力された各測定値を取得する。ただし、これらの予測因子が、他の手段（例えば、ＰｏＣＴや埋め込み型デバイスを用いた手段）により取得されてもよい。また、これらの予測因子として、体液（例えば、涙や汗）の分析結果から類推された値が取得されてもよい。
・基準時刻ｔにおける血清ナトリウム濃度ＳＳＣ（ｔ）の測定値（ｍＥｑ／Ｌ）
・基準時刻ｔにおける血清カリウム濃度ＳＰＣ（ｔ）の測定値（ｍＥｑ／Ｌ）
・基準時刻ｔにおける血清クロール濃度ＳＣＣ（ｔ）の測定値（ｍＥｑ／Ｌ） Specifically, for the following three predictive factors among the seven predictive factors, the medical worker P1 conducts a blood test on the patient and inputs the test results (each measured value) via the operation input unit 140. input. The predictive factor acquisition unit 114 acquires each input measurement value. However, these predictors may be obtained by other means (eg, using PoCT or an implantable device). Further, as these predictive factors, values inferred from analysis results of body fluids (for example, tears and sweat) may be acquired.
・Measurement value of serum sodium concentration SSC (t) at reference time t (mEq/L)
・Measurement value of serum potassium concentration SPC (t) at reference time t (mEq/L)
・Measurement value of serum chloride concentration SCC (t) at reference time t (mEq/L)
　また、以下の３つの予測因子については、医療従事者Ｐ１が、患者に投与する輸液の種類および投与量を決定し、操作入力部１４０を介して決定結果（輸液の種類および投与量）を入力する。予測因子取得部１１４は、入力された輸液の種類および投与量を特定する情報を取得すると共に、サーバ装置２００に格納された輸液情報ＩＤ（図４）を参照して、輸液に含まれるナトリウム量ＩＳＣ（ｍＥｑ／Ｌ）およびカリウム量ＩＰＣ（ｍＥｑ／Ｌ）を取得する。
・基準時刻ｔから将来時刻（ｔ＋Δｔ）までの将来期間Ｔ（０）における輸液投与量ＩＶ（ｍｌ）
・輸液に含まれるナトリウム量ＩＳＣ（ｍＥｑ／Ｌ）
・輸液に含まれるカリウム量ＩＰＣ（ｍＥｑ／Ｌ） Regarding the following three predictive factors, the medical worker P1 determines the type and dose of the infusion to be administered to the patient, and inputs the determination results (type and dose of the infusion) via the operation input unit 140. do. The predictive factor acquisition unit 114 acquires information specifying the type and dosage of the input infusion, and also refers to the infusion information ID (FIG. 4) stored in the server device 200 to determine the amount of sodium contained in the infusion. Obtain ISC (mEq/L) and potassium amount IPC (mEq/L).
- Infusion dose IV (ml) in future period T (0) from reference time t to future time (t + Δt)
・Sodium content ISC (mEq/L) contained in the infusion
・Amount of potassium contained in infusion IPC (mEq/L)
　また、以下の１つの予測因子については、医療従事者Ｐ１が、過去期間Ｔ（－１）における患者の尿量ＵＶを測定し、操作入力部１４０を介して測定結果（尿量ＵＶ）を入力する。予測因子取得部１１４は、入力された尿量ＵＶを取得する。
・過去時刻（ｔ－Δｔ）から基準時刻ｔまでの過去期間Ｔ（－１）における尿量ＵＶ（ｍｌ） Regarding one of the following predictive factors, the medical worker P1 measures the patient's urine volume UV in the past period T(-1) and inputs the measurement result (urine volume UV) via the operation input unit 140. do. The predictive factor acquisition unit 114 acquires the input urine volume UV.
・Urine volume UV (ml) in the past period T (-1) from past time (t-Δt) to reference time t
　なお、予測因子取得部１１４は、院内システム１０または外部ネットワークにおいて提供される電子カルテから予測因子を取得するとしてもよい。 Note that the predictive factor acquisition unit 114 may acquire predictive factors from an electronic medical record provided in the in-hospital system 10 or an external network.
　端末装置１００の血清ナトリウム濃度予測処理部１１１（図３）は、予測の対象の患者について、初回のループであるか否かを判定し（Ｓ３２０）、初回のループである場合には（Ｓ３２０：ＹＥＳ）、後述するＳ３３０の処理をスキップしてＳ３４０の処理に進む。 The serum sodium concentration prediction processing unit 111 (FIG. 3) of the terminal device 100 determines whether or not it is the first loop for the patient to be predicted (S320), and if it is the first loop (S320: YES), the process skips the process of S330, which will be described later, and proceeds to the process of S340.
　次に、端末装置１００の予測実行部１１５（図３）が、血清ナトリウム濃度予測モデルＭＯに、予測の対象の患者について取得された予測因子を入力することにより、将来時刻（ｔ＋Δｔ）における対象者の血清ナトリウム濃度ＳＳＣ（ｔ＋Δｔ）を予測する（Ｓ３４０）。予測実行部１１５は、将来時刻（ｔ＋Δｔ）における対象者の血清ナトリウム濃度ＳＳＣ（ｔ＋Δｔ）の予測結果を示す情報である予測結果データＲＤを生成し、端末装置１００の記憶部１２０に格納する。 Next, the prediction execution unit 115 (FIG. 3) of the terminal device 100 inputs into the serum sodium concentration prediction model MO the predictive factors acquired for the patient to be predicted, thereby predicting the target patient at future time (t+Δt). The serum sodium concentration SSC (t+Δt) is predicted (S340). The prediction execution unit 115 generates prediction result data RD, which is information indicating the prediction result of the subject's serum sodium concentration SSC (t+Δt) at a future time (t+Δt), and stores it in the storage unit 120 of the terminal device 100.
　端末装置１００の予測結果出力部１１６は、予測結果データＲＤに基づき、将来時刻（ｔ＋Δｔ）における対象者の血清ナトリウム濃度ＳＳＣ（ｔ＋Δｔ）の予測結果を出力する（Ｓ３５０）。例えば、予測結果出力部１１６は、予測結果を表示部１３０に表示させる。これにより、医療従事者Ｐ１は、対象の患者に対して、特定の種類の輸液を特定の投与量で投与したときの、将来時刻（ｔ＋Δｔ）における血清ナトリウム濃度ＳＳＣ（ｔ＋Δｔ）の予測値を把握することができる。 The prediction result output unit 116 of the terminal device 100 outputs the prediction result of the subject's serum sodium concentration SSC (t+Δt) at future time (t+Δt) based on the prediction result data RD (S350). For example, the prediction result output unit 116 causes the display unit 130 to display the prediction result. As a result, the medical worker P1 can grasp the predicted value of the serum sodium concentration SSC (t+Δt) at a future time (t+Δt) when a specific type of infusion is administered at a specific dose to the target patient. can do.
　なお、本実施形態では、端末装置１００の報知部１１８（図３）が、対象の患者の血清ナトリウム濃度の予測上昇速度（すなわち、基準時刻ｔにおける血清ナトリウム濃度ＳＳＣ（ｔ）の測定値から、将来時刻（ｔ＋Δｔ）における血清ナトリウム濃度ＳＳＣ（ｔ＋Δｔ）の予測値への濃度上昇速度）が予め設定された閾値（上限値）より速い場合に、所定の方法で報知処理を行う（Ｓ３５０）。このときの血清ナトリウム濃度の上昇速度の閾値は、例えば、ＯＤＳ（浸透圧性脱髄症候群）を回避するために、８～１０ｍＥｑ／Ｌ／日に設定される。また、報知処理の方法としては、例えば、表示部１３０による警報画像の表示や、図示しない音声出力手段による警報音声の出力等が挙げられる。このような報知処理により、医療従事者Ｐ１は、血清ナトリウム濃度の予測上昇速度が過度に速いためにＯＤＳを引き起こすおそれがあることを認識することができる。このような報知処理が行われた場合、例えば医療従事者Ｐ１は、対象の患者に投与する輸液の種類および／または投与量を、血清ナトリウム濃度の予測上昇速度が遅くなるように変更し、端末装置１００に再度Ｓ３１０～Ｓ３５０の処理を実行させる。このような処理を繰り返し実行することにより、医療従事者Ｐ１は、ＯＤＳが引き起こされることを回避しつつ、低ナトリウム血症をできるだけ早く治療できるような、輸液の種類および／または投与量を決定することができる。 In this embodiment, the notification unit 118 (FIG. 3) of the terminal device 100 determines the predicted rate of increase in the serum sodium concentration of the target patient (i.e., from the measured value of the serum sodium concentration SSC(t) at the reference time t). If the rate of increase in serum sodium concentration SSC (t+Δt) to the predicted value at future time (t+Δt) is faster than a preset threshold (upper limit), a notification process is performed using a predetermined method (S350). The threshold value for the rate of increase in serum sodium concentration at this time is set, for example, to 8 to 10 mEq/L/day in order to avoid ODS (osmotic demyelination syndrome). In addition, examples of methods of notification processing include displaying an alarm image on the display unit 130, outputting an alarm sound using an audio output means (not shown), and the like. Through such notification processing, the medical worker P1 can recognize that the predicted increase rate of the serum sodium concentration is too fast and may cause ODS. When such notification processing is performed, for example, the medical worker P1 changes the type and/or dose of the infusion to be administered to the target patient so that the predicted rate of increase in serum sodium concentration will be slower, and The device 100 is caused to execute the processes of S310 to S350 again. By repeatedly performing such a process, the healthcare worker P1 determines the type and/or dosage of the infusion that allows hyponatremia to be treated as quickly as possible while avoiding triggering ODS. be able to.
　その後、血清ナトリウム濃度予測処理が終了されることなく（Ｓ３６０：ＮＯ）、かつ、時間Δｔが経過したとき（Ｓ３７０：ＹＥＳ）、端末装置１００の血清ナトリウム濃度予測処理部１１１は、上述したＳ３１０以降の処理を同様に実行する。２回目以降のループにおいては、基準時刻ｔが、前回のループにおける将来時刻（ｔ＋Δｔ）に更新される。すなわち、例えばＳ３１０においては、予測因子取得部１１４（図３）が、以下の７つの予測因子を取得する。
・基準時刻ｔ（前回のループにおける将来時刻（ｔ＋Δｔ））における血清ナトリウム濃度ＳＳＣ（ｔ）の測定値（ｍＥｑ／Ｌ）
・基準時刻ｔ（同）における血清カリウム濃度ＳＰＣ（ｔ）の測定値（ｍＥｑ／Ｌ）
・基準時刻ｔ（同）における血清クロール濃度ＳＣＣ（ｔ）の測定値（ｍＥｑ／Ｌ）
・将来期間Ｔ（０）（前回のループにおける将来期間Ｔ（０）に続くさらに将来の期間）における輸液投与量ＩＶ（ｍｌ）
・輸液に含まれるナトリウム量ＩＳＣ（ｍＥｑ／Ｌ）
・輸液に含まれるカリウム量ＩＰＣ（ｍＥｑ／Ｌ）
・過去期間Ｔ（－１）（前回のループにおける将来期間Ｔ（０））における尿量ＵＶ（ｍｌ） Thereafter, when the serum sodium concentration prediction process is not terminated (S360: NO) and the time Δt has elapsed (S370: YES), the serum sodium concentration prediction processing unit 111 of the terminal device 100 performs the process after S310 described above. Execute the same process. In the second and subsequent loops, the reference time t is updated to the future time (t+Δt) in the previous loop. That is, for example, in S310, the predictive factor acquisition unit 114 (FIG. 3) acquires the following seven predictive factors.
・Measurement value (mEq/L) of serum sodium concentration SSC (t) at reference time t (future time (t+Δt) in the previous loop)
・Measurement value (mEq/L) of serum potassium concentration SPC (t) at reference time t (same)
・Measurement value (mEq/L) of serum chloride concentration SCC (t) at reference time t (same)
・Infusion dose IV (ml) in future period T(0) (further future period following future period T(0) in the previous loop)
・Sodium content ISC (mEq/L) contained in the infusion
・Amount of potassium contained in infusion IPC (mEq/L)
・Urine volume UV (ml) in past period T(-1) (future period T(0) in previous loop)
　また、２回目以降のループ（Ｓ３２０：ＮＯ）においては、端末装置１００のモデル更新部１１７（図３）が、前回のループにおける予測因子と、今回のループにおける基準時刻ｔ（前回のループにおける将来時刻（ｔ＋Δｔ））における血清ナトリウム濃度ＳＳＣ（ｔ）の測定値と、が対応付けられたデータを含む訓練データＴＤを用いた機械学習によって、血清ナトリウム濃度予測モデルＭＯを更新する（Ｓ３３０）。これにより、血清ナトリウム濃度予測モデルＭＯが、対象の患者の特性（病態、体質等）により適合したモデルとなり、血清ナトリウム濃度の予測精度が向上する。 In addition, in the second and subsequent loops (S320: NO), the model update unit 117 (FIG. 3) of the terminal device 100 uses the predictive factors in the previous loop and the reference time t in the current loop (in the future in the previous loop). The serum sodium concentration prediction model MO is updated by machine learning using training data TD including data associated with the measured value of serum sodium concentration SSC(t) at time (t+Δt) (S330). Thereby, the serum sodium concentration prediction model MO becomes a model that is more suitable for the characteristics (pathological condition, constitution, etc.) of the target patient, and the prediction accuracy of the serum sodium concentration is improved.
　以上のような処理が繰り返し実行されている中で、血清ナトリウム濃度予測処理の終了指示があったとき（Ｓ３６０：ＮＯ）、血清ナトリウム濃度予測処理は終了する。 While the above-described processes are being repeatedly executed, when there is an instruction to end the serum sodium concentration prediction process (S360: NO), the serum sodium concentration prediction process ends.
Ａ－５．実施例：
　上述した血清ナトリウム濃度予測モデルＭＯの実施例について、以下説明する。本実施例では、名古屋大学医学部附属病院　糖尿病・内分泌内科に入院した低ナトリウム血症の患者（１６例）の治療過程で得られたデータを訓練データＴＤとして用いた機械学習により、血清ナトリウム濃度予測モデルＭＯを作成した。 A-5. Example:
An example of the above-mentioned serum sodium concentration prediction model MO will be described below. In this example, serum sodium concentration was predicted by machine learning using data obtained during the treatment process of hyponatremic patients (16 cases) admitted to the Department of Diabetes and Endocrinology, Nagoya University Hospital. A model MO was created.
　図７は、本実施例の血清ナトリウム濃度予測モデルＭＯの予測精度を示す説明図である。図７には、１３３個の観測点について、血清ナトリウム濃度の測定値と、血清ナトリウム濃度予測モデルＭＯによる血清ナトリウム濃度の予測値との関係が示されている。図７に示すように、各観測点は、概ね、完全な予測を示す直線の近辺に分布しており、非常に高い予測精度を実現していると言える。なお、図７に示す結果は、学習モデルとして線形サポートベクター回帰を用い、検証方法として１０分割交差検証を行ったものである。 FIG. 7 is an explanatory diagram showing the prediction accuracy of the serum sodium concentration prediction model MO of this example. FIG. 7 shows the relationship between the measured values of serum sodium concentration and the predicted values of serum sodium concentration by the serum sodium concentration prediction model MO for 133 observation points. As shown in FIG. 7, the observation points are generally distributed near a straight line that indicates perfect prediction, and it can be said that very high prediction accuracy has been achieved. Note that the results shown in FIG. 7 are obtained by using linear support vector regression as a learning model and performing 10-fold cross validation as a verification method.
　図８は、学習モデルおよび予測因子を種々変更した場合における血清ナトリウム濃度予測モデルＭＯの予測精度を示す説明図である。図８には、学習モデルおよび予測因子の各組合せについて、血清ナトリウム濃度予測モデルＭＯの予測精度を表す２つの評価指標（ＲＭＳＥおよびＲ２）の値が示されている。学習モデルの選択肢としては、（１）線形回帰、（２）線形サポートベクター回帰、（３）バギング木の３つが設定されている。また、予測因子の選択肢としては、上述した予測因子の７つの項目から取捨選択した８種類の組合せが設定されている。図８には、予測因子の７つの項目が符号（図１参照）により示されている。図８において、黒丸が付された項目が予測因子として採用された項目であり、黒丸が付されていない項目が予測因子として採用されなかった項目である。 FIG. 8 is an explanatory diagram showing the prediction accuracy of the serum sodium concentration prediction model MO when the learning model and prediction factors are variously changed. FIG. 8 shows the values of two evaluation indices (RMSE and R 2 ) representing the prediction accuracy of the serum sodium concentration prediction model MO for each combination of the learning model and the prediction factor. Three learning model options are set: (1) linear regression, (2) linear support vector regression, and (3) bagging tree. In addition, eight types of combinations selected from the seven predictor factor items described above are set as predictor options. In FIG. 8, seven items of predictive factors are indicated by symbols (see FIG. 1). In FIG. 8, items with black circles are items that were adopted as predictors, and items without black circles are items that were not adopted as predictors.
　図８に示すように、学習モデルおよび予測因子の各組合せのいずれについても、概ね高い予測精度を示した。学習モデルについては、線形回帰および線形サポートベクター回帰を採用した場合に、特に予測精度が高くなった。予測因子については、上記実施形態のように、７つの項目すべてを予測因子として採用した場合が最も予測精度が高くなるが、７つの項目のうちのいずれか１つまたは２つを省略しても、予測精度の低下は最小限であった。そのため、各医療機関の設備や運用の違いにより、予測因子の構成が多少異なっても、血清ナトリウム濃度予測モデルＭＯを用いることによって高精度に血清ナトリウム濃度を予測できると言える。 As shown in FIG. 8, each combination of learning model and predictor showed generally high prediction accuracy. Regarding the learning model, the prediction accuracy was particularly high when linear regression and linear support vector regression were adopted. As for the predictors, as in the above embodiment, the prediction accuracy is highest when all seven items are adopted as predictors, but even if one or two of the seven items are omitted, , the decrease in prediction accuracy was minimal. Therefore, it can be said that even if the composition of the predictive factors differs somewhat due to differences in the equipment and operation of each medical institution, the serum sodium concentration can be predicted with high accuracy by using the serum sodium concentration prediction model MO.
Ａ－６．本実施形態の効果：
　以上説明したように、本実施形態の端末装置１００は、将来時刻（ｔ＋Δｔ）における血清ナトリウム濃度を予測するための情報処理装置であって、予測因子取得部１１４と、予測実行部１１５とを備える。予測因子取得部１１４は、予測の対象者について、基準時刻ｔにおける血清ナトリウム濃度ＳＳＣ（ｔ）の測定値と、基準時刻ｔから将来時刻（ｔ＋Δｔ）までの期間である将来期間Ｔ（０）におけるナトリウム摂取量を表す指標値と、を含む予測因子を取得する。予測実行部１１５は、予測因子と将来時刻（ｔ＋Δｔ）における血清ナトリウム濃度ＳＳＣ（ｔ＋Δｔ）の測定値とが対応付けられた訓練データＴＤを用いた機械学習により生成された血清ナトリウム濃度予測モデルＭＯに、対象者について取得された予測因子を入力することにより、将来時刻（ｔ＋Δｔ）における対象者の血清ナトリウム濃度ＳＳＣ（ｔ＋Δｔ）を予測する。 A-6. Effects of this embodiment:
As described above, the terminal device 100 of the present embodiment is an information processing device for predicting serum sodium concentration at a future time (t+Δt), and includes a predictive factor acquisition unit 114 and a prediction execution unit 115. . The predictive factor acquisition unit 114 acquires the measured value of the serum sodium concentration SSC(t) at the reference time t and the measured value of the serum sodium concentration SSC(t) at the reference time t and the future period T(0), which is the period from the reference time t to the future time (t+Δt), for the person to be predicted. Obtain an index value representing sodium intake and a predictive factor including the index value. The prediction execution unit 115 uses a serum sodium concentration prediction model MO generated by machine learning using training data TD in which a prediction factor is associated with a measured value of serum sodium concentration SSC (t+Δt) at a future time (t+Δt). , the serum sodium concentration SSC (t+Δt) of the subject at a future time (t+Δt) is predicted by inputting the predictive factors obtained for the subject.
　このように、本実施形態の端末装置１００では、予測の対象者についての、基準時刻ｔにおける血清ナトリウム濃度ＳＳＣ（ｔ）の測定値と、将来期間Ｔ（０）におけるナトリウム摂取量を表す指標値と、を含む予測因子を、血清ナトリウム濃度予測モデルＭＯに入力することにより、将来時刻（ｔ＋Δｔ）における対象者の血清ナトリウム濃度ＳＳＣ（ｔ＋Δｔ）を精度良く予測することができる。そのため、本実施形態の端末装置１００によれば、予測の対象者についての上記予測因子を取得するだけで、将来時刻（ｔ＋Δｔ）における対象者の血清ナトリウム濃度ＳＳＣ（ｔ＋Δｔ）を精度良く予測することができる。これにより、例えば、医療従事者Ｐ１は、ＯＤＳが引き起こされることを回避しつつ、低ナトリウム血症をできるだけ早く治療できるような、輸液の種類および／または投与量を決定することができる。また、ＩＣＵ治療期間の予測に基づく入室最適化を実現することができる。 As described above, in the terminal device 100 of the present embodiment, the measured value of the serum sodium concentration SSC(t) at the reference time t and the index value representing the sodium intake in the future period T(0) for the person to be predicted. By inputting the predictive factors including , into the serum sodium concentration prediction model MO, it is possible to accurately predict the subject's serum sodium concentration SSC (t+Δt) at future time (t+Δt). Therefore, according to the terminal device 100 of the present embodiment, it is possible to accurately predict the serum sodium concentration SSC (t+Δt) of a subject at a future time (t+Δt) by simply acquiring the above-mentioned predictive factors for the subject of prediction. Can be done. This allows, for example, the healthcare worker P1 to decide on the type and/or dosage of the infusion that will allow the hyponatremia to be treated as soon as possible while avoiding triggering ODS. Furthermore, optimization of admission to the ICU based on prediction of the ICU treatment period can be realized.
　また、本実施形態の端末装置１００は、さらに、対象者の血清ナトリウム濃度の予測結果を出力する予測結果出力部１１６を備える。そのため、本実施形態の端末装置１００によれば、装置の使用者に、将来時刻（ｔ＋Δｔ）における対象者の血清ナトリウム濃度ＳＳＣ（ｔ＋Δｔ）の予測値を把握させることができる。 Furthermore, the terminal device 100 of this embodiment further includes a prediction result output unit 116 that outputs a prediction result of the subject's serum sodium concentration. Therefore, according to the terminal device 100 of this embodiment, the user of the device can grasp the predicted value of the subject's serum sodium concentration SSC (t+Δt) at a future time (t+Δt).
　また、本実施形態の端末装置１００は、さらに、訓練データＴＤを取得する訓練データ取得部１１２と、訓練データＴＤを用いた機械学習によって血清ナトリウム濃度予測モデルＭＯを作成するモデル取得部１１３とを備える。そのため、本実施形態の端末装置１００によれば、他の装置を用いずとも血清ナトリウム濃度予測モデルＭＯを取得することができ、該モデルを用いて将来時刻（ｔ＋Δｔ）における対象者の血清ナトリウム濃度ＳＳＣ（ｔ＋Δｔ）の予測を実行することができる。 Furthermore, the terminal device 100 of the present embodiment further includes a training data acquisition unit 112 that acquires training data TD, and a model acquisition unit 113 that creates a serum sodium concentration prediction model MO by machine learning using the training data TD. Be prepared. Therefore, according to the terminal device 100 of the present embodiment, the serum sodium concentration prediction model MO can be obtained without using any other device, and the serum sodium concentration of the subject at future time (t+Δt) can be calculated using this model. A prediction of SSC(t+Δt) can be performed.
　また、本実施形態の端末装置１００は、さらに、対象者についての予測因子と将来時刻（ｔ＋Δｔ）における血清ナトリウム濃度ＳＳＣ（ｔ＋Δｔ）の測定値とが対応付けられたデータを含む更新された訓練データＴＤを用いた機械学習によって、血清ナトリウム濃度予測モデルＭＯを更新するモデル更新部１１７を備える。そのため、本実施形態の端末装置１００によれば、血清ナトリウム濃度予測モデルＭＯを、対象者の特性により適合したモデルとすることができ、血清ナトリウム濃度の予測精度を向上させることができる。 In addition, the terminal device 100 of the present embodiment further includes updated training data that includes data in which predictive factors for the subject are associated with measured values of serum sodium concentration SSC (t+Δt) at future time (t+Δt). A model updating unit 117 is provided that updates the serum sodium concentration prediction model MO by machine learning using TD. Therefore, according to the terminal device 100 of the present embodiment, the serum sodium concentration prediction model MO can be a model that is more suitable for the characteristics of the subject, and the accuracy of predicting the serum sodium concentration can be improved.
　また、本実施形態の端末装置１００では、予測因子取得部１１４は、対象者が摂取する摂取物（輸液および／または飲水）の種類を特定する情報を取得すると共に、各摂取物に含まれるナトリウム量を示す情報（輸液情報ＩＤ）を参照して、将来期間Ｔ（０）におけるナトリウム摂取量を表す指標値を取得する。そのため、本実施形態の端末装置１００によれば、使用者が摂取物（輸液）の種類を指定するだけで、将来期間Ｔ（０）におけるナトリウム摂取量を表す指標値の取得を実現することができ、より効率的に血清ナトリウム濃度の予測を実行することができる。 Furthermore, in the terminal device 100 of the present embodiment, the predictive factor acquisition unit 114 acquires information specifying the type of ingestion (infusion fluid and/or drinking water) ingested by the subject, and the sodium content in each intake. An index value representing the amount of sodium intake in the future period T(0) is obtained by referring to information indicating the amount (infusion information ID). Therefore, according to the terminal device 100 of the present embodiment, it is possible to realize the acquisition of the index value representing the sodium intake in the future period T(0) simply by the user specifying the type of ingestion (infusion). The prediction of serum sodium concentration can be performed more efficiently.
　また、本実施形態の端末装置１００では、予測因子は、さらに、将来期間Ｔ（０）における水分摂取量を表す指標値と、過去時刻（ｔ－Δｔ）から基準時刻ｔまでの期間である過去期間Ｔ（－１）における水分排出量を表す指標値と、を含む。そのため、本実施形態の端末装置１００によれば、血清ナトリウム濃度の予測精度を向上させることができる。 Furthermore, in the terminal device 100 of the present embodiment, the predictive factors further include an index value representing the amount of water intake in the future period T(0) and a past period that is the period from the past time (t-Δt) to the reference time t. and an index value representing the amount of water discharged during period T(-1). Therefore, according to the terminal device 100 of this embodiment, the prediction accuracy of serum sodium concentration can be improved.
　また、本実施形態の端末装置１００では、予測の対象者は、低ナトリウム血症の患者である。そのため、本実施形態の端末装置１００によれば、低ナトリウム血症の患者に対して所定の治療を施したときの血清ナトリウム濃度の予測を実現することができる。 Furthermore, in the terminal device 100 of this embodiment, the subject of prediction is a patient with hyponatremia. Therefore, according to the terminal device 100 of this embodiment, it is possible to predict the serum sodium concentration when a predetermined treatment is given to a hyponatremic patient.
　また、本実施形態の端末装置１００は、さらに、基準時刻ｔにおける対象者の血清ナトリウム濃度ＳＳＣ（ｔ）の測定値から、将来時刻（ｔ＋Δｔ）における対象者の血清ナトリウム濃度ＳＳＣ（ｔ＋Δｔ）の予測値への濃度上昇速度が予め設定された閾値より速い場合に、所定の方法で報知する報知部１１８を備える。そのため、本実施形態の端末装置１００によれば、低ナトリウム血症の患者の治療において、血清ナトリウム濃度の上昇速度が過度に速くなることを抑制することができ、例えばＯＤＳの発生を回避することができる。 Furthermore, the terminal device 100 of the present embodiment further predicts the subject's serum sodium concentration SSC(t+Δt) at a future time (t+Δt) from the measured value of the subject's serum sodium concentration SSC(t) at the reference time t. The apparatus includes a notification section 118 that notifies you in a predetermined manner when the rate of increase in concentration to this value is faster than a preset threshold. Therefore, according to the terminal device 100 of the present embodiment, in treating a patient with hyponatremia, it is possible to suppress the rate of rise in serum sodium concentration from becoming excessively fast, and for example, to avoid the occurrence of ODS. I can do it.
Ｂ．変形例：
　本明細書で開示される技術は、上述の実施形態に限られるものではなく、その要旨を逸脱しない範囲において種々の形態に変形することができ、例えば次のような変形も可能である。 B. Variant:
The technology disclosed in this specification is not limited to the above-described embodiments, and can be modified into various forms without departing from the gist thereof. For example, the following modifications are also possible.
　上記実施形態における端末装置１００の構成は、あくまで一例であり、種々変形可能である。また、上記実施形態における血清ナトリウム濃度予測モデル取得処理および血清ナトリウム濃度予測処理の内容は、あくまで一例であり、種々変形可能である。例えば、上記実施形態では、端末装置１００が、血清ナトリウム濃度予測モデルＭＯを作成することによって血清ナトリウム濃度予測モデルＭＯを取得しているが、端末装置１００が、他の装置（例えば、院内システム１０内のサーバ装置２００や外部ネットワーク上の装置）により生成された血清ナトリウム濃度予測モデルＭＯを取得するとしてもよい。この場合には、端末装置１００が訓練データ取得部１１２を有する必要はない。 The configuration of the terminal device 100 in the above embodiment is just an example, and can be modified in various ways. Further, the contents of the serum sodium concentration prediction model acquisition process and the serum sodium concentration prediction process in the above embodiment are merely examples, and can be modified in various ways. For example, in the embodiment described above, the terminal device 100 obtains the serum sodium concentration prediction model MO by creating the serum sodium concentration prediction model MO. The serum sodium concentration prediction model MO generated by the internal server device 200 or a device on an external network may be acquired. In this case, the terminal device 100 does not need to have the training data acquisition unit 112.
　上記実施形態では、血清ナトリウム濃度予測モデルＭＯの更新（図６のＳ３３０）が実行されているが、血清ナトリウム濃度予測モデルＭＯの更新が実行されなくてもよい。この場合には、端末装置１００がモデル更新部１１７を有する必要はない。 In the above embodiment, the serum sodium concentration prediction model MO is updated (S330 in FIG. 6), but the serum sodium concentration prediction model MO does not need to be updated. In this case, it is not necessary for the terminal device 100 to have the model updating section 117.
　上記実施形態では、報知処理（図６のＳ３５０）が実行されているが、報知処理が実行されなくてもよい。この場合には、端末装置１００が報知部１１８を有する必要はない。 In the above embodiment, the notification process (S350 in FIG. 6) is executed, but the notification process does not need to be executed. In this case, the terminal device 100 does not need to have the notification section 118.
　上記実施形態における血清ナトリウム濃度予測モデルＭＯの作成に用いられる予測因子（図１）は、あくまで一例であり、種々変形可能である。例えば、上記実施形態では、予測因子が、基準時刻ｔにおける血清カリウム濃度ＳＰＣ（ｔ）の測定値、基準時刻ｔにおける血清クロール濃度ＳＣＣ（ｔ）の測定値、および、輸液に含まれるカリウム量ＩＰＣを含んでいるが、予測因子がこれらの内の少なくとも１つを含まなくてもよい。また、上記実施形態では、過去期間Ｔ（－１）における水分排出量を表す指標値として、尿量ＵＶが用いられているが、これに代えて、あるいは、これに加えて、他の指標値（例えば、体重変化量、血漿浸透圧、血糖値等）が用いられてもよい。また、上記実施形態では、将来期間Ｔ（０）におけるナトリウム摂取量を表す指標値として、将来期間Ｔ（０）における輸液投与量ＩＶおよび／または輸液に含まれるナトリウム量ＩＳＣが用いられているが、これに代えて、あるいは、これに加えて、他の指標値（例えば、食べ物等の経口により摂取される物の摂取量および／または該物に含まれるナトリウム量等）が用いられてもよい。 The predictive factors (FIG. 1) used to create the serum sodium concentration prediction model MO in the above embodiment are merely examples, and can be modified in various ways. For example, in the above embodiment, the predictive factors are the measured value of the serum potassium concentration SPC(t) at the reference time t, the measured value of the serum chloride concentration SCC(t) at the reference time t, and the potassium amount IPC contained in the infusion. However, the predictor may not include at least one of these. Further, in the above embodiment, the urine volume UV is used as an index value representing the amount of water excreted in the past period T (-1), but instead of or in addition to this, other index values may be used. (For example, body weight change, plasma osmolarity, blood sugar level, etc.) may be used. Furthermore, in the above embodiment, the infusion dose IV and/or the sodium amount ISC contained in the infusion in the future period T(0) is used as an index value representing the sodium intake in the future period T(0). , Instead of this, or in addition to this, other index values (for example, the intake amount of something ingested orally such as food and/or the amount of sodium contained in the object, etc.) may be used. .
　上記実施形態では、医療従事者Ｐ１が使用する端末装置１００が、予測因子取得部１１４および予測実行部１１５を含む血清ナトリウム濃度予測処理部１１１を備えているが、血清ナトリウム濃度予測処理部１１１の少なくとも一部の機能が端末装置１００ではなく、他の装置（例えば、院内システム１０内のサーバ装置２００や外部ネットワーク上の装置）内に存在していてもよい。例えば、血清ナトリウム濃度予測処理部１１１の少なくとも一部の機能が、電子カルテシステムの一機能として組み込まれていてもよい。このような形態とすれば、本明細書に開示される技術を、例えばＤｏｃｔｏｒ－ｔｏ－Ｄｏｃｔｏｒの遠隔医療コンサルテーションに利用することもできる。なお、このような形態においては、端末装置１００および／または該他の装置が、特許請求の範囲における情報処理装置（または情報処理システム）の一例となる。また、血清ナトリウム濃度以外に、他の体内電解質濃度（例えばカリウム濃度）等を予測し、血清ナトリウム濃度の予測値と共に出力する装置であってもよい。 In the above embodiment, the terminal device 100 used by the medical worker P1 includes the serum sodium concentration prediction processing section 111 including the prediction factor acquisition section 114 and the prediction execution section 115. At least some of the functions may exist not in the terminal device 100 but in another device (for example, the server device 200 in the in-hospital system 10 or a device on an external network). For example, at least part of the functions of the serum sodium concentration prediction processing section 111 may be incorporated as a function of the electronic medical record system. With such a configuration, the technology disclosed in this specification can be used for, for example, doctor-to-doctor remote medical consultation. In addition, in such a form, the terminal device 100 and/or the other device are an example of an information processing device (or information processing system) in the scope of the claims. Further, in addition to serum sodium concentration, the device may predict other electrolyte concentrations (for example, potassium concentration) in the body and output them together with the predicted value of serum sodium concentration.
　上記実施形態では、低ナトリウム血症の患者に対して所定の治療を施したときの血清ナトリウム濃度を予測するための情報処理を例示しているが、本明細書に開示される技術は、これに限られず、他の場面における血清ナトリウム濃度の予測にも同様に適用可能である。例えば、本明細書に開示される技術は、長距離ランナーといった運動をしている人を対象として、経口による補給を行ったときの将来時刻における血清ナトリウム濃度を予測する際にも同様に適用可能である。 The above embodiment exemplifies information processing for predicting serum sodium concentration when a predetermined treatment is given to a patient with hyponatremia. The present invention is not limited to, but can be similarly applied to prediction of serum sodium concentration in other situations. For example, the technology disclosed herein can be similarly applied to predicting the serum sodium concentration at a future time when oral supplementation is performed in an athletic person such as a long-distance runner. It is.
　上記実施形態において、ハードウェアによって実現されている構成の一部をソフトウェアに置き換えるようにしてもよく、反対に、ソフトウェアによって実現されている構成の一部をハードウェアに置き換えるようにしてもよい。 In the above embodiments, a part of the configuration realized by hardware may be replaced with software, or conversely, a part of the configuration realized by software may be replaced by hardware.
１０：院内システム　１００：端末装置　１１０：制御部　１１１：血清ナトリウム濃度予測処理部　１１２：訓練データ取得部　１１３：モデル取得部　１１４：予測因子取得部　１１５：予測実行部　１１６：予測結果出力部　１１７：モデル更新部　１１８：報知部　１２０：記憶部　１３０：表示部　１４０：操作入力部　１５０：インターフェース部　１９０：バス　２００：サーバ装置 10: Hospital system 100: Terminal device 110: Control unit 111: Serum sodium concentration prediction processing unit 112: Training data acquisition unit 113: Model acquisition unit 114: Predictor acquisition unit 115: Prediction execution unit 116: Prediction result output unit 117: Model update section 118: Notification section 120: Storage section 130: Display section 140: Operation input section 150: Interface section 190: Bus 200: Server device

Claims (10)

  1. 　将来時刻における血清ナトリウム濃度を予測するための情報処理装置であって、
    　前記予測の対象者について、
    　　基準時刻における血清ナトリウム濃度の測定値と、
    　　前記基準時刻から前記将来時刻までの期間である将来期間におけるナトリウム摂取量を表す指標値と、
    を含む予測因子を取得する予測因子取得部と、
    　前記予測因子と前記将来時刻における血清ナトリウム濃度の測定値とが対応付けられた訓練データを用いた機械学習により生成された血清ナトリウム濃度予測モデルに、前記対象者について取得された前記予測因子を入力することにより、前記将来時刻における前記対象者の血清ナトリウム濃度を予測する予測実行部と、
    を備える、情報処理装置。     An information processing device for predicting serum sodium concentration at a future time, comprising:
    Regarding the target of the prediction,
    a measured value of serum sodium concentration at a reference time;
    an index value representing sodium intake in a future period that is a period from the reference time to the future time;
    a predictor acquisition unit that acquires a predictor including
    Inputting the predictive factor obtained for the subject into a serum sodium concentration prediction model generated by machine learning using training data in which the predictive factor and the measured value of serum sodium concentration at the future time are associated. a prediction execution unit that predicts the serum sodium concentration of the subject at the future time;
    An information processing device comprising:
  2. 　請求項１に記載の情報処理装置であって、さらに、
    　前記対象者の血清ナトリウム濃度の予測結果を出力する予測結果出力部を備える、情報処理装置。     The information processing device according to claim 1, further comprising:
    An information processing device comprising a prediction result output unit that outputs a prediction result of the serum sodium concentration of the subject.
  3. 　請求項１または請求項２に記載の情報処理装置であって、さらに、
    　前記訓練データを取得する訓練データ取得部と、
    　前記訓練データを用いた前記機械学習によって前記血清ナトリウム濃度予測モデルを作成するモデル取得部と、
    を備える、情報処理装置。     The information processing device according to claim 1 or 2, further comprising:
    a training data acquisition unit that acquires the training data;
    a model acquisition unit that creates the serum sodium concentration prediction model by the machine learning using the training data;
    An information processing device comprising:
  4. 　請求項１または請求項２に記載の情報処理装置であって、さらに、
    　前記対象者についての前記予測因子と前記将来時刻における血清ナトリウム濃度の測定値とが対応付けられたデータを含む更新された前記訓練データを用いた前記機械学習によって、前記血清ナトリウム濃度予測モデルを更新するモデル更新部を備える、情報処理装置。     The information processing device according to claim 1 or 2, further comprising:
    The serum sodium concentration prediction model is updated by the machine learning using the updated training data that includes data in which the predictive factor for the subject is associated with the measured value of serum sodium concentration at the future time. An information processing device comprising a model update unit that updates a model.
  5. 　請求項１または請求項２に記載の情報処理装置であって、
    　前記予測因子取得部は、前記対象者が摂取する摂取物の種類を特定する情報を取得すると共に、各前記摂取物に含まれるナトリウム量を示す情報を参照して、前記将来期間におけるナトリウム摂取量を表す指標値を取得する、情報処理装置。     The information processing device according to claim 1 or 2,
    The predictive factor acquisition unit acquires information specifying the type of intake that the subject ingests, and also refers to information indicating the amount of sodium contained in each intake to determine the sodium intake in the future period. An information processing device that obtains an index value representing.
  6. 　請求項１または請求項２に記載の情報処理装置であって、
    　前記予測因子は、さらに、
    　　前記将来期間における水分摂取量を表す指標値と、
    　　過去時刻から前記基準時刻までの期間である過去期間における水分排出量を表す指標値と、
    を含む、情報処理装置。     The information processing device according to claim 1 or 2,
    The predictor further includes:
    an index value representing water intake in the future period;
    an index value representing the amount of water discharged in a past period that is a period from a past time to the reference time;
    Information processing equipment, including.
  7. 　請求項１または請求項２に記載の情報処理装置であって、
    　前記対象者は、低ナトリウム血症の患者である、情報処理装置。     The information processing device according to claim 1 or 2,
    The information processing device, wherein the subject is a hyponatremic patient.
  8. 　請求項７に記載の情報処理装置であって、さらに、
    　前記基準時刻における前記対象者の血清ナトリウム濃度の測定値から、前記将来時刻における前記対象者の血清ナトリウム濃度の予測値への濃度上昇速度が予め設定された閾値より速い場合に、所定の方法で報知する報知部を備える、情報処理装置。     The information processing device according to claim 7, further comprising:
    in a predetermined method when the concentration increase rate from the measured value of the serum sodium concentration of the subject at the reference time to the predicted value of the serum sodium concentration of the subject at the future time is faster than a preset threshold. An information processing device including a notification unit that provides notification.
  9. 　将来時刻における血清ナトリウム濃度を予測するための情報処理方法であって、
    　前記予測の対象者について、
    　　基準時刻における血清ナトリウム濃度の測定値と、
    　　前記基準時刻から前記将来時刻までの期間である将来期間におけるナトリウム摂取量を表す指標値と、
    を含む予測因子を取得する工程と、
    　前記予測因子と前記将来時刻における血清ナトリウム濃度の測定値とが対応付けられた訓練データを用いた機械学習により生成された血清ナトリウム濃度予測モデルに、前記対象者について取得された前記予測因子を入力することにより、前記将来時刻における前記対象者の血清ナトリウム濃度を予測する工程と、
    を備える、情報処理方法。     An information processing method for predicting serum sodium concentration at a future time, the method comprising:
    Regarding the target of the prediction,
    a measured value of serum sodium concentration at a reference time;
    an index value representing sodium intake in a future period that is a period from the reference time to the future time;
    obtaining a predictive factor comprising;
    Inputting the predictive factor obtained for the subject into a serum sodium concentration prediction model generated by machine learning using training data in which the predictive factor and the measured value of serum sodium concentration at the future time are associated. predicting the serum sodium concentration of the subject at the future time by;
    An information processing method comprising:
  10. 　将来時刻における血清ナトリウム濃度を予測するためのコンピュータプログラムであって、
    　コンピュータに、
    　前記予測の対象者について、
    　　基準時刻における血清ナトリウム濃度の測定値と、
    　　前記基準時刻から前記将来時刻までの期間である将来期間におけるナトリウム摂取量を表す指標値と、
    を含む予測因子を取得する処理と、
    　前記予測因子と前記将来時刻における血清ナトリウム濃度の測定値とが対応付けられた訓練データを用いた機械学習により生成された血清ナトリウム濃度予測モデルに、前記対象者について取得された前記予測因子を入力することにより、前記将来時刻における前記対象者の血清ナトリウム濃度を予測する処理と、
    を実行させる、コンピュータプログラム。     A computer program for predicting serum sodium concentration at a future time, the computer program comprising:
    to the computer,
    Regarding the target of the prediction,
    a measured value of serum sodium concentration at a reference time;
    an index value representing sodium intake in a future period that is a period from the reference time to the future time;
    a process of obtaining a predictor including
    Inputting the predictive factor obtained for the subject into a serum sodium concentration prediction model generated by machine learning using training data in which the predictive factor and the measured value of serum sodium concentration at the future time are associated. A process of predicting the serum sodium concentration of the subject at the future time by;
    A computer program that runs
PCT/JP2023/016397 2022-05-12 2023-04-26 Information processing device, information processing method, and computer program WO2023218957A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2022-078809 2022-05-12
JP2022078809 2022-05-12

Publications (1)

Publication Number Publication Date
WO2023218957A1 true WO2023218957A1 (en) 2023-11-16

Family

ID=88730404

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2023/016397 WO2023218957A1 (en) 2022-05-12 2023-04-26 Information processing device, information processing method, and computer program

Country Status (1)

Country Link
WO (1) WO2023218957A1 (en)

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070088333A1 (en) * 2005-10-13 2007-04-19 G&L Consulting, Llc Method and system for infusing an osmotic solute into a patient and providing feedback control of the infusing rate
JP2020074769A (en) * 2018-11-01 2020-05-21 花王株式会社 Method for preparing nucleic acid derived from skin cells of subject

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070088333A1 (en) * 2005-10-13 2007-04-19 G&L Consulting, Llc Method and system for infusing an osmotic solute into a patient and providing feedback control of the infusing rate
JP2020074769A (en) * 2018-11-01 2020-05-21 花王株式会社 Method for preparing nucleic acid derived from skin cells of subject

Similar Documents

Publication Publication Date Title
Kent et al. Device closure of patent foramen ovale after stroke: pooled analysis of completed randomized trials
Kamstrup et al. Genetically elevated lipoprotein (a) and increased risk of myocardial infarction
JP6526735B2 (en) Methods and systems for predicting changes in physiological and clinical conditions
Robinson et al. Predictors of relapse following response from a first episode of schizophrenia or schizoaffective disorder
JP6828055B2 (en) Estimating and using clinician assessment of patient urgency
JP6018623B2 (en) Staged alarm method for patient monitoring
JP6298454B2 (en) Method for evaluating hemodynamic instability index indicator information
Duncan et al. Identification of a novel mutation in the CDHR1 gene in a family with recessive retinal degeneration
US20110137852A1 (en) Sepsis Monitoring and Control
Sutin et al. Psychological distress, self-beliefs, and risk of cognitive impairment and dementia
JP6072021B2 (en) Evaluation system and evaluation method
Tran et al. Hypertension in children
Neleman et al. FFR-guided PCI optimization directed by high-definition IVUS versus standard of care: the FFR REACT trial
JP2017503532A (en) Clinical decision support system based on triage decision
JP2023546866A (en) Systems and methods for providing clinical decision support
Bell et al. “Smoker’s Paradox” in patients treated for severe injuries: lower risk of mortality after trauma observed in current smokers
Luo et al. Blood leukocyte counts in Alzheimer disease
Feigal et al. Associations between positive emotional well-being and stress-induced myocardial ischemia: Well-being scores predict exercise-induced ischemia
Koami et al. Thromboelastometric analysis of the risk factors for return of spontaneous circulation in adult patients with out-of-hospital cardiac arrest
JP2020187687A (en) Medical information processing device
Fishman Historical evolution in the understanding of Stargardt macular dystrophy
WO2023218957A1 (en) Information processing device, information processing method, and computer program
Alghothani et al. The effect of glycemic variability on counterregulatory hormone responses to hypoglycemia in young children and adolescents with type 1 diabetes
Siegal et al. Evaluating anticoagulant strategies for atrial fibrillation in patients with cancer: challenges and opportunities
TW202107478A (en) Computing device, portable device and computer-implemented method for predicting major adverse cardiovascular events

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 23803436

Country of ref document: EP

Kind code of ref document: A1

DPE1 Request for preliminary examination filed after expiration of 19th month from priority date (pct application filed from 20040101)